DeSciNYC

DeSciNYC

← Back to videos

Talk & transcript

Prophetic AI

Eric Wollberg

Transcript

Download text

Automatically generated transcript. May contain errors. Select a timestamp to watch that moment on YouTube.

0:05

[Music] [Music]

0:28

do okay um thank you guys for coming um I know it's like a cold day but this is the 13th D NYC which is amazing so it has been it has been a journey I think we keep going we keep growing which is kind of cool um I don't know if that's because D andc is growing or because interest in D Community is growing or New York is growing I'm not super sure uh but either way thank you for coming um during the first couple ones we we just sort of jumped right in and people there was a lot of people here who are like uh what is

1:10

decentralized science so I thought I would give like a uh like like a few minute primer for those people who are here for who here who here is for nft NYC week one two okay actually not that many I think all the nft NC people are at the uh some sort of ape party thing you mind that your head when was there um yeah in the nft I think there's a more intellectual crowd here uh who here knows what D is okay a couple more what what what what made other people interested here lucid dreaming what lucid dreaming

1:53

okay cool cool okay awesome um well thank you guys for coming I will get like a little um overview around um what like why why I think decentralized science is really important so like today the government and like uh controls a lot of how research is shaped so they they control the funding they control uh who works at different universities and imagine a world where instead of like the government shaping research and shaping uh academics and shaping funding like the it was the academics were shaped by like the internet and like the internet was like the biggest laboratory in the

2:34

world and it wasn't like a university which is sort of exclusive to only admit certain types of people that's how I know we're in New York yeah we're definitely in New York um so basic basically what's happening is like we're putting much too we're putting way too much uh trust into these large institutions and what happens when when you do it and there's that single point of failure is sort of all of that uh the the same ideas sort of get propagated so the the things that were being funded 30 years ago are just sort of repeatedly funded over and over again and one of the promises of decentralized science is that uh more people can have ideas

3:14

around uh types of stuff that they want to see in the world and also more people can contribute to them so right now like what does science look like it looks like things that are in academic institution that have been funded by the government that have people that have been admitted to PHD programs to um work on that sort of research and um the promise of D is is that uh that incentives can be more broadly spread so that anyone who wants to participate can in the same way you can anyone can buy a Bitcoin and there's not sort of as much uh regulation on who does it anyone can participate in a science in a science thing in the same way without having to opt in to do all this formal paperwork before and uh I think uh the time is now

3:57

for this to happen and like the desent Iz movement is taking off and we're very close to having this happen which is very exciting so um uh and I I think one other thing I really want to impress upon is the importance of uh incentives and how incentives work in science in science right now the incentives in academics are sort of uh misconstrued because you have to uh get accepted a paper and then that paper gives you status and then that status will allow you to apply for a gr uh professorship and then that professorship ship allows you to get a grant and then the grant allows you to exit and be an adviser to some other company but the point is is like there's so many uh levels of

4:38

indirection here that it's better to just have a much more direct system where people that are doing interesting research or people that are doing have interesting ideas are able to uh get uh data and get uh those sort of newer ideas out and the the way to make that happen is not through trying to change academic institutions but it's try and think about what are the incentives models in there and how can you replicate those so that uh people can probably make more money pushing their ideas using a design method as opposed to going to work for a university where someone makes like 180 a year oh so um this is a little table I made

5:18

around I'm calling it trai versus D um so who here knows like traditional F trai is like Trad and um and defi is like decentralized Finance so Trad side we're just going to start at the top I don't like talking about funding because basically if you're funding if you're not getting money in the traditional model it's not because the funding sucks it's probably because you suck and so um I they don't me to start with it um there but it it is there to to start with and um you know the trai method relies on uh government institutions and grants whereas the dii method re re relies on like decentralized crypto

5:58

financing use box style crowdsourcing um the publication method is also very different um so we have like in the Trin trai method you have like peerreview journals and everything's behind the pay wall and there's like a lengthy process to uh get things accepted and in the design method it's uh articles are shorter everything has Open Access you have to pay for the things you want and it's much more granular um peer review in the traditional method is anonymous which leads to all sorts of pervert weird incentives where you don't know who's judging you and they have no the the bigger problem is they have no incentive to do a good job so like if I'm

6:40

peer-reviewing something most people when they peer review they're trying to like finish that peerreview article as quickly as possible because if they do a good or a bad job there's no incentive but in the def in the D way of doing it um everything is like a staking and a gambling thing where if you predict accurately uh the type of paper that will be successful in a few years you stand to make more money which I think is like a much better and better way of doing it because now the person who's reviewing it is actually going to care a lot more um data sharing is you know data sharing in academic institutions is often times very siloed or it's uh you have to email them uh in the D method you can just post your data on chain with an open format and you have to pay

7:21

to get access to it and when you pay to get access to a data set it it's not a bad thing it's actually sort of a good thing because it incentivizes people to uh maintain high quality data sets um Community this is one of the bigger ones um I think this is I think there's a lot more scientists out there in the world than are in than we than we typically think of today so today we think of a scientist as someone who's maybe at an R&D lab or uh at at a university but in the same that's like sort of thinking about like programmers as people who sort of majored in computer science but the the number of programmers out there is actually probably much broader right there's a lot of self-taught people there's a lot of people who are just genuinely interested in who might have a

8:02

9 to5 who think of themselves as scientists or programmers in the outside and so this is like a really cool thing of trying to expand what it means to be a scientist and also not necessarily think of a scientist as the person who does every part of the experiment right like when someone participates in a clinical trial they're actually part being scientists that are participating in science and thinking of that as a in a thinking of Science in a broader sense and something that I think is going to be very cool you guys can swing around if you want our people to the game wait there's there's room in the back hello hello hey hey hello so many people um yeah we're just talking around

8:42

science and Desi okay um so uh governance is another really interesting thing is like how does governance work today in traditional science it works by you know the pi makes all the decisions but you can imagine like governance can be much more community-driven and um um and you know it's not won't be as Locked In by institutional thinking when there's anyone with a token can vote um accessibility and is one uh and reproducibility is another one like right now uh there's a there is a huge reproducibility crisis in science and I think one of the one of the most important reasons why I'm so uh bullish on uh dii is not because DSi is so

9:24

powerful but because the the bar is so low for traditional science because like most science if you try and replicate a paper or something like that it can't be replicated at all um you can't get the data set you you can't reproduce the methods so you're it's basically means that you're building on top of something that's very very hard to build on top of whereas in the in the D version you can imagine everything being uh in a Jupiter notebook uh easily oneclick reproducible have the AI check that it's reproducible and so it makes it'll make it things much it'll make it'll make it so that you can build on top of Science and a much easier way um so this is about incentives so

10:07

incentives Drive progress so if we if we change the incentives uh we will change it the things that are important in a dite project are token issuance so each project can have its own currency which is really cool so you think of like a research project not necessarily as a research project but as a as a package like a company almost and each person that participates is a is a partial owner of it so say someone contributes data to a science to a research project they get to own a little bit of a little part of that data a little part of the IP that's created from that project or someone say someone does analysis or say someone promotes the project um and finds and finds people that can help so it's thinking of a project like like a

10:49

company almost and then giving everyone that participates in that company a stake in it so they're incentivized it's like stock options they're incentivized to do a really good job with it where as in traditional thing the there people are incentivized but uh you know you'll see a lot of in instances where uh pis put pressure on people to publish research and that causes them to make up data but if the incentives were align we would hope that maybe that wouldn't be the case um each each project can have a pool of data which will be very cool um types of types of types of data pools that I'm very excited about um are things like like longevity so right now

11:31

like longevity is so popular right like everyone Brian who here knows who Brian Johnson is wow lots and lots of people um like right now like PE there's this notion that maybe it is possible to significantly extend our life past just 100 years maybe to 150 200 and eventually even longer um and so but the problem is is we don't know if any of these things work right so we need to figure out a way to decentralize the testing of these types of things and I think dite is like a really opport a really perfect place for that where um people can participate in these studies and contribute their data and then own whatever partially own whatever type of IP is Created from that oring is another

12:13

one sleep Research heart rate variability and genomic data are other ones that I would be very excited about um data quality uh you can have an exchange so once you imagine that uh dii is a thing you can go into a website and just trade participate in figuring out which type of things you want to participate in and just like in the same way coin Market or uh coinbase is there you can just like go in and pick and choose the types of dnip projects that you're really excited about um experiments will change instead of think of experiments as a just a binary yes no you can think of them as like a prediction market so that you could participate and like oh I think this experiment is actually going to work very well or the experiment is not

12:55

and I'm willing to stake a certain amount of money on it okay so that was sort of the D siize field does anyone have any questions on that someone must have one yes like notbook that exist no but that's a good idea there should be some uh Jupiter notebooks that would help with that for sure yeah Hippa Hippa I mean Hippa is only Avail only important if you're a covered provider and so this goes outside of that where it instead instead of a hospital participating in dite you would go directly to the patients where they wouldn't be subject to the same

13:36

thing how do you expect to get genomic data it's like just me uh I mean I would in in the perfect world I've thought about it it was like you would you would use something like 23 in meat but with a bit more data privacy and you would then uh get your data and then you would upload it and then people would pay to act like you would have some sort of AI scan across it and look at the types of things that give a broad understanding of your genomics and then or some sort of program and then people would pay to access certain parts of your data and then you would get that would be the incentive for you to do that yeah um the tokens that you imagine

14:18

what would would they have any fundamental value or or like rights to IP or or Revenue share or I mean this would be dependent on the project obviously but I think one things that would make sense are IP um governance so you can participate in in the project in certain types of ways um and and presumably block things what and presumably block things maybe too yeah exactly exactly yeah you can say this this we are going to allow this type of data this type of research to go forward this type of research to not go forward everything would be sort of a vote um weighted in in certain types of ways and um yeah so that'll be cool yeah how do you value different

14:59

give contributions to the project so that's like set in stone before the project is is before you participate so when you participate someone puts up a bounty for if you do this you will be rewarded this much and then based on if that price matches your expectation you you get it so maybe someone contributing data is more or less valuable than someone contributing anal and data analysis but that'll just be dependent on the project but I think what's cool is you'll know from the start of it like it little be sort of on on chain or on contract knowing for how much it is yeah goad anyone else yeah um what about in the end everything

15:43

has everything yeah kind of a big underground clal trials movement out of the hands of big F you know there a lot of very and they're not going to wait 10 years yeah so I mean the group even like this this side of the gr gets theu

16:19

they yeah so I mean that that's so let's do it I mean yeah and I think what's cool is now you've you created an incentive model for people more people to participate in it um and yeah I I agree and if you go if you go to other places in the world like uh Bermuda or there's like Nicaragua or places they're starting to do these sort of more advanced clinical trials so it it's it's a matter of time yeah resarch um yeah

17:01

go um and there's there's another if you go in the back in the yellow bag there's another one that looks like let me get um yeah so I mean one of the big ones people are looking into uh like data sharing so people are looking to share data sets with each other um people are looking to do a life extension people are looking to do cognitive stuff so for like brain development I think a lot of I think one of the issues that is a little bit disappointing at the moment is a lot of these are funding things so they're just they're just doing the funding they're not doing so much onchain research and I think the real the real promise of this is not is I mean the funding is okay but the real promise in dii lies with um building a

17:42

community and and participating in research in a in a decentralized way and not just like crowdsourcing money and then um yeah right yeah to have like High Fidelity clinical trials or work that you're doing like you let's say that you're getting blood drawn you'll go and you'll do it at uh institution where they're monitoring you and they're doing things to hold things constant yeah if you have everybody really doing it in a distributed system how do you actually make sure that things are held constant so that you have like high quality data yeah it just changes the data set so it changes the variance on it and that's just a new factor that you and you hope that because there even if there's higher variance it'll be okay because uh there's a larger data

18:23

set yeah um so unfortunately in the past we've seen networks and systems that were supposed to be designed as decentralized that due to eff poorly thought outness and human nature ended up becoming while still decentralized Network and system have centralized Powers over time uh what could uh we put into these service systems to make sure we don't end up with a technically decentralized network but end up accidentally centralizing all of our data or uh the hierarchy of gets approved yeah I mean I think I think Randomness is a is a more is a better way of doing that so like I think you're

19:04

describing as like the failure of markets which is basically like a monopoly and that happens when there's just one or two institutions doing something and um this will and but if you have like enough competition um that doesn't really become an issue and so I think you would just hope that if there's more people doing it it's not as big of a deal okay anyone else one question oh yeah with with this Marketplace token holders would be active participants and potentially speculators yeah so how do you prevent against the latter warping the market speculators yeah well I mean it's like the stock

19:45

market like I don't remember there's that saying like in the short term it does something but over the long term it's like a weighing mechanism and I think it's the same thing like even over the long term you'd expect for the ver the truth to come out in the market and for the speculation to just be noise right but more recent examples in crypto for example speculation I feel like has been playing a larger role especially with this me how do you insulate dii against that I don't think that that's necessarily a bad thing because I think it generates excitement but look what happens with mean points they go up very quickly and then they go down very quickly that doesn't really that that sort of actually kind of works in some ways right like that's what you that's what

20:26

you want like there's all this this and then the down but over the course of 6 months like nobody really made besides the people who participated in the Ponzi scheme no one actually made money right and so that's that's kind of a good thing cool okay okay wait I have a couple more things that I okay so yes so now I will talk about D NYC okay so um D NYC this is our 13th one um we very clearly in New York um if you want to find out our past videos we have a website now called dnyc uh there is a QR code and it has videos it has other things there

21:07

too um okay this is the cool part so we were very obstinate about checking making sure everyone checked in today and the reason being is because we're going to do like a dite token and there's going to be like a fair airdrop this is just like a very Community Based project but everyone that like was checked in who who here knows what an airdrop is okay a lot of people so we're going to we're going to based on the email address that's B that's with your registered person we're going to give everyone 100 tokens um each time someone comes to a dite thing and we're going to try and decentralize D NYC and you know so this is like everyone now is going to be like the first uh holder of this D NYC token

21:48

which will be very cool um each time you participate and you come again to another event um you'll get more tokens um people that tweet out uh around it will also get more tokens people that help at the door will also get more tokens and the idea and everyone that volunteers gets tokens and the idea is we're trying to like distribute these D NTC tokens uh very fair oh and speakers speakers and host of course sorry oh actually I have a kick for both of you I I did actually I remember I I forgotten it the last three times but this time I actually did remember it um um so and uh of someone you're done speaking Yeah so the speakers and the the hosts anyone that participates we're trying to spread these tokens out as as as uh as fairly

22:29

as possible and the idea is um so if you find a speaker or if you're volunteer you you get more tokens and is to like spread these tokens out and then eventually maybe do a research project or do something else with them but sort of the the key point is each person that comes today we're going to like just go through the email the checkin that's why you're so careful about it and make sure who's checked in and we'll and you'll be there'll be a follow-up email uh with that and if you oh if you tweet about this also you're spreading the word or you something like that or you tag us on Instagram you also get tokens so we're trying to give away as many tokens as possible here what's the what's the what's the token supp uh we haven't even figured that out yet it's just points right now but it's TVD we going to what

23:11

set yeah of course or maybe I don't mean I someone else would probably should do that but all right now it's just around tokens what's the what's the thicker symbol uh d e s c i D NYC will be the ticker but so you'll get an email with more information around your token allocation and how to claim it oh we're also going to put a thing where you have to claim it within 30 days because we don't want tokens to just sort of be floating around indefinitely because that makes it hard to figure out uh the governance model of this so it's sort of a use it or lose it type thing you have to go in claim it and then take the tokens back with

23:52

you um yeah this is what I just said you'll have we'll send out an email you'll have 30 days days to get it um okay some other things around D NYC so that's very exciting because now we're trying to like decentralize d NYC and I think this is the first time that that's ever been done which will be very exciting as well um we have a telegram group uh this was sort of dead for a little bit but we're trying to bring it back now um if you want to participate in dii after the event ends or before the next event begins um oh we have a lot of people joining also thank you to the Safeguard bought for uh we were getting spam for a little bit Yeah so everyone can participate post your pictures there um yeah go get

24:35

in there okay I'm glad I opened it up before it was really closed and stuck and uh stff but now it's uh more open I'll probably edit this part out of the video I don't know we'll see anyone else still dinging or are we good it's all right well wait I agree what it's a good proof for for what like you should keep it oh yeah yeah yeah okay yeah it's the link what the link is link ah interesting good to know they just updated

25:15

I okay I think we I think we dinged out there okay so this is uh oh this is our past speakers we've had many many speakers we've had uh people talking about drugs nfts research uh okay okay so um here's the other thing you guys are going to be excited about the next event is here we've scheduled it on um May 16th so if you want to uh sh in the back uh if you want to participate in that um you can take a you can sign up there we don't have an active thing or you can go to Luma D

25:56

nc14 good okay and then oh this is important okay so it's as you can see there's a lot of people here right so um it's not really Poss oh who if you were a volunteer raise your hand so let's give a round of applause to all the people they're the ones who are and the even if you're not wearing the white coat you had the white coat so you can uh give a round of applause and if you would like to participate as a volunteer next time and be in out of the kindness of your own heart or because you're into uh D NYC tokens uh you can scan this scan this link um it would be immensely grateful to anyone who is helping out um and it's

26:37

a lot of fun too we'll get dinner you'll get to meet each other it'll be a bit more closed you can get to meet each other a little bit more which will be exciting so that's the that's the link here and it's a very short form you just fill out your email and stuff like that everyone still want it okay okay okay so that's all um let's do a quick group picture um of everyone here turn off TR it that way yeah that's what um do you guys want to how are we going to be in

27:18

[Music] it get in there if you can't see the camera the camera can't see you

27:37

amazing okay looks good okay and now um without further Ado you want to well let's give our speaker a big round of applause before before we have we have another token holder uh thank you um to uh uh so our wonderful host here who you explain all right so first off that creature in your slides is going to give me Nightmare so I need your help it's a t grade bro it's a t grade so my name is Manu uh welcome to Arch this is our office uh we are a crypto lending company primarily we also do some other stuff also host events uh we've been

28:18

around for a couple years um feel free to check us out Arch ling.com I want to keep you from uh this man here so Eric thanks thank you

28:37

who's sitting on the [Music] remot oh a final ask for um

28:56

people um make sure to take pictures um today um and tag us on like uh tag us on Instagram or Twitter D NYC it really helps uh we'll retweet you it'll be amazing but you actually have to do it because sometimes I feel like I have to like nudge people a few times but take out your phone like actually do it yes pull it out pull it out I'm going to be I'm going to be annoying about it and it'll it'll help bring more people and it'll make it a more fun event for everyone here so it's it it's a collective thing we all participate we all do our part It's it's not that big of an Ask come on and and take a picture

29:36

of the surroundings whatever and and tag us on Instagram or Twitter D NYC talk tell me after that you did it or raise your hand if you did it now did you do it did you actually do it it's like this isn't like when you the dentist asks you if you floss this is like this is like an important question not like a aing question okay I'm going to annoy you again later so you'll have you did it I will you will no it's not now when you have to do it [Music] now okay okay without further Ado

30:17

Eric um good even um first you Michael thank you for inviting me m thank you for hosting us and thank you all for coming um in 2018 I was uh sitting outside the old city of Jerusalem I was living there at the time um and I was sitting in this very you know beautiful Park under olive trees I was reading a lot of theology as one does when you live in Jerusalem and I noticed something kind of very compelling for me which was that whether it's Abraham Muhammad or Buddha these prophets received their prophetic wisdom in their dreams and engraved on the interior of this device is uh

30:58

our our motto I guess you can say and it says Prometheus stole fire from the gods and we will steal dreams from the prophets now Prometheus right is the Western myth in terms of you know talking about technology right um now my co-founder pointed this out we did not uh discover fire right if lightning hits uh you know uh dry grass you have fire what we learned to do was control it and with the control of fire look at what we've done we built modern industrial civilization um and then fundamentally changed you know our reality so then I I begs the question um what would you do

31:40

you we stole fire from the gods what if we stole dreams from the prophets now we've been dreaming forever I mean animals dream octopi dream uh you know mammals dream Etc um we've been dreaming for you know millions of years um what if we could learn to control that um now by the way people do do this right 55% of people self-report having had a lucid dream at least once in their life maybe I'll ask the audience who's at a Luci dream awesome a little more than 55% that's a selection effect I can imagine that um now a smaller percentage of people appear to have a natural prity for it I'm part of that second lucky group so I had my first s dream when I

32:20

was about 12 years old I woke up and I was equal parts terrified and excited right I had never heard anyone talk about lucid dreaming Hagrid did not pick me up on my 11th birthday so like what the [ __ ] going on right um now I found these early internet forums talking about this thing newed dreaming so first of all s relief like other people do this fantastic um and then I got really really into it in college I didn't really care for going to class and so what I did was I uh you know I I I looked at the work of a gentleman named Dr Steven Lear now Dr Steven Lear okay did his PhD at Stanford in 1980 um and and and created among many things routines and behaviors that you could utilize to actually increase your capacity for Lucidity things like Reality Checking dream journaling

33:02

pneumonic induction which is really a bad idea by the way like it's like you falling asleep and being like I'm going to be in a dream the next thing is going to be in a dream and you just stay awake it actually doesn't work um but but I did get myself to a point where I was having kind of two lucid dreams a week and my absolute Zenith um and then I I graduated college um and entered the workforce and it's impossible to maintain that level of like kind of you know rigor um and so in 200 18 I I I kind of you know noticed this this peculiarity though I'm not sure how peculiar it really is um and I decided Well what what if we could actually create a device that could induce this brain State on demand now in 2018 I I

33:42

found two independent research groups 2013 2014 using electrical stimulation both alternating and direct current stimulation and they were actually successful to a certain extent but uh you know Lucidity is a scale you can be aware that you're in a dream but not in control for example now they got intuition and and something called disassociation which is a very interesting thing for some you know the ability to Pivot between looking at things from the first or third person noticeably absent was controlled now electrical stimulation um internally atep prophetic we call it the vacuum tubes of non-invasive neuros stimulation um three reasons you know I've just completely blown class this so we might not even use this um uh besides you guys

34:24

probably you're good with slides um now electrical simulation Three core reasons why it's a very limited technology okay first is depth your skull was evolved to keep [ __ ] away from your brain so getting electricity p through skull is quite difficult now should you achieve this what happens is a phenomenon simply called spread and spreads across the surface of it which means that you don't have precision and you don't have the ability to steer it now the reason that they were successful at all is that the neural corelates of lucid dreaming is REM with the prefrontal cortex activated that this makes a ton of sense um REM is when you're in a dream I think we all know that your prefrontal cortex is where your conscious awareness decision-making working memory are um

35:06

it's why you feel as present as you do and I hope you do right now um and so imagine merging those two things together naturally it's like a plus Bal C right so the reason electrical simulation was successful at all albeit limitedly is that they were still simulating the the surface of the prefrontal cortex during R so it was having some effect but electrical stimulation as I've already mentioned is an extremely Limited technology it's not the thing that's going to get you the level of accuracy generalizability um and precision that you need so in 2018 I had to do something very hard you know I had to look I was investing actually in deep Tech at the time and if you invest in deep Tech what you really are trying to determine right is like is this in research or is it in development because if it's in research that's not really

35:46

what the asset class of venture capital was designed to do um and so I had to table it you know I I've done many other things since um but something awesome happened in 2022 I found two two core technologies that underg good what we do so the first our Nur simulation modality is trans cranial Focus ultrasound now if electrical stimulation is the integr uh is the the vacuum tube let me assure you that tfus is the integrated circuit back to those three p uh three previous points okay um depth centimeters into the brain no problem Precision millimeters you're going from no Precision to millimeters that's a paradigm shift and

36:28

then three and perhaps just as important as the Precision is the ability to steer it in three dimensions your brain fires in threedimensional neural firing patterns right so you want something that can do that so I found TF us uh being used by a gentlem named Dr Jason gy uh at the University of Arizona he was using it to induce deep meditative States whether you had ever meditated a day in your life drop you into a meditative state that might take somebody a decade of practice to achieve now this this caught my eye I was super excited and I also found that it was being being used for Focus positive mood it's being looked at for addiction Alzheimer's uh you know uh inoperable brain tumors if you increase the

37:08

frequency it's being used to BL inoperable brain tumors and actually very very cool this is not really related at all but like to to to to my company but a couple of months ago they published a paper where they took somebody who was quote unquote minimally conscious and buzzed the rain with TF us 8 days after that they were walking we built the modern you know technological error on the integrated circuit if if tfus is the integrated circuit for non-invasive neuros simulation think about the world that we are going to build because what we're talking about is not interfacing with electrons and and a screen we're talking about something that's inducing conscious experience and everything from the cave paintings you know in Spain to

37:50

Alexander weeping you know because he had no more worlds to conquer to what we're experiencing right now was a conscious experience so the profundity of this is non-trivial um core technology number two a neural Transformer now to understand the importance of this let me explain how tfus which has been utilized for neurom modulation since 2004 let me explain how it's being used in the State ofthe art what they do is they take an fmri probably many um and they look for the areas of statistical significance something called rois and you know a brain state has many areas of of statistical significance but what they do is they use some very statistical mechanics and maybe something that I really doubt any of us

38:31

especially people who work in ml would call ml um and they look at the area of the most the most statistical significance and then they take a single element trans transducer or a multi-element transducer let me explain that really quickly an ultrasonic transducer is a ceramic with a p Electric piece of material or Crystal when you run electricity through that it oscillates creating the ultrasonic wave so they take uh either a single element transducer or a multielement transducer really acting as a single element transducer and puls it at that area of most statistical significance for 2 minutes 5 minutes 10 minutes 20 minutes Etc and uh you know this is kind of a kin and by the way it's had an extraordinary success like this is It's

39:12

Already Done un amazing things um but it's like playing one note on one instrument and expecting a song now the neural Transformer opens up a very very different world for this technology and the importance of this is that first of all neuroimaging data is very unique now we use both EG and fmri data it's a multimodal neural Transformer that we built um if you look at FM um you know it's enormous and very unique spatial data you know EEG is is unique in the context of its temporality um so it's very unique data and it's insanely information dense okay the frame rate or the sampling rate I should say of fmri is 2.1 seconds so

39:52

every 2.1 seconds you're getting an inordinate amount of spatial data about this brain state and we use simultaneous EGF from our data which means in 2.1 seconds you take 2100 samples of EG it's insanely information dense now this is a very good thing for us for building very accurate models with less data than let's say an llm but it required an architecture that could really you like do anything interesting with this so um okay how do we put these two core Technologies you know together okay so this is a Halo um we designed this first of all with card 79 um they did they designed nuring for Elon when Max hodak left uh nurlink actually we have to cut that out I'm not sure that's public I'm

40:34

going to pause um but anyway they worked in nerling um and uh and and so you know we couldn't ask for a better partner they don't normally you know work with companies as as as young as us um but they were really excited about what we're doing so we're really appreciative um and so okay headband four channels EG this will go up but right now it's four channels EG two ultrasound transducers plac temporally okay temporality of uh is very important it's the thinnest part of your skull it allows us to do Cross Beam stimulation to cover the full prefrontal cortex which is our Target region and um it will also be a way in which we calibrate this that I can't talk about but um okay so how does this work um what's your name Andre Andre

41:17

yeah okay Andre right okay so Andre goes to sleep he puts this on and I I give to you put it on um and you go to sleep the first thing that is important to no is the the EEG on the headband that is the thing that is prompting the neural Transformer um it is prompting the neural Transformer with your current brain state right so the first thing it's looking for is is Andre in rem once Andre is in rem the model triggers the tfus it sends out the first pulse of tfus and then what the model is looking for is the EG signature of Lucidity gamma frequency spikes during Rim so it basically goes is Andre in R is he in r is he in rem oh [ __ ] is he in

41:58

rem is he Lucid is he Lucid okay he's Lucid let's keep him Lucid let's keep him Lucid he's not in rem anymore let's turn it off that's how this works uh we we showcased by the way about two months ago our first neural Transformer model Morpheus one the world's first multimodal neural Transformer um we've done a lot of things you can if you're on Twitter you can follow our Twitter one of the key things that I really believe in when you're doing something as audacious as as this and I'm not I know I know this is [ __ ] crazy um but what you do is you have to show your work and so if you follow us on Twitter uh really that's a only place we post I don't even have an Instagram um you know we we we every week you'll see us making kind of Headway doing something new Etc whatever

42:39

but one thing that I will say that I'm very excited by is um about next week um maybe if some things get held up in customs you know the week after we will start neuros stimulation um Three core experiences that are going to be on this device when it launches focus and positive mood these are already things that have been done with TF us we're just replicating and then three of course is lucid dreaming um so I don't want to like I really want to hear your questions and your thoughts but um I I I really want to leave you with with this which is I I can't actually really you know impress upon you further how profound this is I maybe you maybe you think so already but you know this is here inevitable my company May Fail many companies fail

43:20

most companies fail but tfus and neurot Transformers th those are just beginning this is the worst neurot Transformers will ever be just imagine you know we talk about dii okay neuroimaging data is extremely hard to you know to come by it's costly to aggregate we do reinforcement learning on the headband with the EG it's also how we kind of personalize the model to you over time Etc but it also allows us to aggregate insane quantities of EG data which isn't great but actually because it's it's paired with the the the perational tfus it's actually something much more akin to almost synthetic feron um and so you know we're going to like

44:01

be releasing a piece hopefully you know by the end of the month or or next month that really talks about the vision of where this technology is going and and what and the role we aim to play I I'll tease it for you here um when is this coming out tomorrow [ __ ] okay well you can keep it in I don't really care it's fine um everyone right now is very afraid of this you know this concept of of the AI takeoff scenario right um it really does appear right that like AI is kind of improving markedly at a rapid rate um and we don't really know what's going to happen there and there's a lot of anxiety um you know one of my favorite movies is her I don't know how many people have seen her I'm about to ruin the ending so sorry about that at

44:43

the end of her the OS uh played by Scarlet Johansson and Theodore um she tells him that she's leaving and he's like well where are you going now you have to understand they've been experiencing an AI takeoff scenario in that movie right she's getting more more and more intelligent she's doing nonverbal non symbolic communication Etc whatever and then eventually she's like listen like your your world is very slow and so we're leaving and he says where are you going and she says it'd be hard to explain but if you ever get there come find me nothing will ever you know take us apart you know he seems to like accept this at the end of the movie and so on but I was obsessed by this like where

45:24

the [ __ ] did she go I want to go like you know and you kind of think about the anxiety here it's like I don't know if you guys have dogs remember but like you know when you leave a dog in your in their apartment and you're going out and it's looking at you with extraordinary pain right it's like where the [ __ ] are you going are you going to come back Etc that I think is a very analogous thing to what we feel in this context of theyi not only right like we thought we were the gods right now but we're afraid we're the dogs you know um that's lame I'm sorry about that um but but what I offer you is an alternative to this um um what I offer you is a qualia takeoff scenario now think about this any one of our ancestors 500 years ago um had a

46:05

very small life relative to the number of conscious experiences that they would have compared to us it's trivia this is why stained glass windows were very popular in Gothic Europe right a stain glass window is like the most psychedelic [ __ ] those people ever saw right it's it's no no no I mean a good idea right to associate it with God if you're the church now over time we already are on a quality takeoff scenario right radio synthetic you know synthetic uh you know hearing or or or kind of uh noise generation television okay screens okay then you have you know the internet now you have geni and now we have this and and I think the thing that I'm most interested in and and so on and and most kind of compelled by is to make

46:47

sure that we can have a qualia takeoff scenario alongside this AI take off scenario it's going to be [ __ ] weird I'm not going to lie to you like I I genuinely like I I I'd be lying to you if if there wasn't like some level of like concern or or like kind of like fear that I have like you know even as a person like who's like envisioned this and and is working with a team of course to build you know this is just non [ __ ] trivial but it's here and so you can't in the same way kind of run away from AI like you can't run away from this and so what I kind of hope for us is that you know we increasingly have more and more you know variable and interesting conscious experiences you know when the AI takes all of our jobs or [ __ ] like that like maybe that's what we do we

47:27

just like experience new things and novel things together um and and and and hopefully we find where Samantha went all right I'll stop you okay perfect now we will now we will Grill if you want okay okay I have a question um what is the underlying mechanism by which it works like are you TF us yeah yeah are you are you increasing blood flow are you heating up a part of the brain are you like what is the so yeah so TF us right trans cranial Focus ultrasound what you're doing is you're transferring um via ultrasound

48:10

mechanical energy which is making the ion channels and so so forth open up just enough so you get enough just enough potassium or not whatever to get the neuron to fire so that's actually the mechanism of of action of tfus is it heating up the brain under uh under so not uh not at the levels that we're doing this so I'm sure one of the questions that people are going to ask is like oh are you going to have to go through an FDA approval process the answer is no um we don't make medical claims um and uh you know uh we're well below the the predetermined thresholds which is 720 m per CM squar that still is very low below the thermal band of tfus so I mentioned that um just last year they did the first higher intensity

48:51

right um TF us surgery on an inoperable brain tumor right they were able to oblate an inoperable range that would have almost certainly killed that person uh with tfus uh that that's a much much higher frequency of tfus got it okay um I have a question Ben go ahead can you doing with with tfus yeah so trans cranial gopper that you absolutely and actually the way that we would calibrate it would involve uh some some the use of the tfus for for for kind of calibrating this I can't speak more know further than that um so yeah it's kind of like echolocation um you know I very cool um do you think you could actually use the technology in the

49:32

future to like break ups for like sh so the same technology that I just mentioned right for that's ablation um so very different than what we're doing right I get that but but but absolutely um so yeah okay I see but your thesis is you stimulated at such a low hurt to promote certain neuron transmitters to fire it it just makes the neuron fire so so right you're transferring the mechanical energy through to the neuron that activates it makes it fire and the patterns of those activations are the things that in the aggregate create a brain State and to be very clear I mean this is already been again been done for 20

50:13

years this is in like a a thesis focus and positive mood already been done I found it again through somebody who's getting in induc deep meditative State they never meditated a day in their life um so yeah important what's the level of precision when you're targeting in 3D space like is it down to I know nuring talks about like being able to read a level of individual Norm but like what that's not necessary right for this right so you can do millimeters you can do cimet Etc depending on you know the resolution you want what we want to do right is we we train the model on right we have I didn't mention this we have a collaboration with the ders Institute which is the top lucid dreaming lab in the world in the world the Rabon

50:54

University it's led by Dr Martin drer who discovered is actually the guy who discovered the neuroc correlates for lucid dreaming we do four fmis a week um with you know Dutch undergraduate students and now kind of expert Lo dreamers um so you take that Mo that data as well as open source data um and we train the model with it right and the model then right uh prompting that model right it it's creating kind of you know sequences kind of I mentioned this you know analogy of it's like playing you know the old Paradigm of TF us it's like playing one note on one instrument and expecting a song we created Orchestra because we use M you know multielement transducers and we use the model to guide in space and time the pulses to recreate that brain State um and so the

51:35

resolution that we need is we're just trying to create a cacophony of of of of of of preal Cortex activity in the context of lucid freeming whereas for example positive mood is very specific you're targeting the dorsal lalal CeX so um it depends and by the way uh one thing I'm sorry uh one thing to say is with tfus and the 3D thre dimensional steering you can actually create like like acoustic so you can like let's say I don't think we we need this but let's say okay you're charging the dorsal L PR the cortex imagine creating a little 3D acoustic hologram that stimulates it precis um like does that answer your question what do you mean um okay uh okay next person who's

52:17

tweeted about us does anyone tweeted internet sucks here so I have it you want to go yeah just simple question for those of us who may haven't had one you define a lucid dream that would probably a good idea so um uh okay a regular dream right REM your prefontal cortex is not active you're kind of in this like persistent state of Alzheimer right that's kind of why like you you just kind of accept the crazy [ __ ] you know that's happening so again uh you know prefontal cortex activation you are there you are aware in your dream I mean this is best culturally explained right in the Inception um with the key difference that you're not being not able to loose a dream you know with people there are

52:59

people whove come up to me and said oh yeah no I have I have I'm not open I will keep an open mind but I I I I I doubt it as of now um and so what I'm describing is it is so you're aware in your dream imagine how consciously you are right now that in your dream and what the limiting factor that experience is what you imagine because what you imagine becomes but maybe let me actually explain a little a couple of things of of like the uses of this right that probably be also help thank you for the question but so you know there are three kind of core use cases um you know Recreation it's the ultimate BR experience right um You can fly um you can do something else that begins with

53:39

an N like you can do a lot of things right you can make buildings you know appear out of the ground you can talk to your dream characters which is very interesting and by the way very similar to an llm in the sense that sometimes you ask them something and they say something like extremely profound and then you ask them something and they say something absolutely you know gibberish um the second is productive I mean we have investors who code in their elusive dreams run the code in the morning um the guy who designed the first render of our Hardware is a fashionate hardware designer that designed in his dreams the the history of dreaming and Discovery is quite long right Irwin schinger did his physics in his dreams um savador Aly painted in his dreams uh the structure of the Benzene molecule was discovered a dream KLA dream famously called it's a snake eating its tail the benzine

54:20

molecules of CC ring carbon atoms It's actually an OP okay there we go thank you be honest um and and sasan ramanujan is probably my favorite example right he sends a book of of an infinite series to GE Hardy and they invite him over to to England and they're like this is unbelievable like where are the proofs and he says oh no like there are no proofs like you know Shiva gaps me in my dreams so the history of that is very long and then the third thing is metaphysical um when asked you know why I I like love Luc dreaming and what it's been for me it's actually less about the experience itself which is incredible um it's about how it reenchant the waking

55:02

world and if I had to kind of look around in society and broadly say like what the [ __ ] is wrong because obviously something is it's that you know through secularization and a variety of other things we disenchanted our world but our world is extremely interesting and actually very enchanting and I think you know lucid dreaming is something that's really powerful for that I'll give you a concrete example on the metaphysical side um if you go to the subreddit for Lu dreaming it's one of the largest subreddits in the world um you'll see what people are doing there's over half million people and I found this guy and it was really like really stuck with me you know he had a bad relationship with his father his father died he never had to have the convers he never got to have the conversation that he wished he [ __ ] had with this guy and so in a

55:43

lucid dream he had that conversation because when you know somebody very well and you have them as a dream character you have a mental representation of how they would respond to things probably not perfect but it's close um that's really [ __ ] incredible um so those are the use cases I guess but there's probably many more there was a question here sure um it was a public session yeah so this is this primar can this be used to help sleep and is there going to be like an app so you can monitor to see if you're in the r or yes yes yes um you know I I had all these slides and I just threw out the script my apologies we have an app um we're building it um it'll have two kind of

56:24

sides so classic kind of whoop or a Met tracking so on so forth your question on sleep um tfus has been used actually to induce slow wave uh sleep um it's already been done um that would probably require an FDA approval process something that we're not I'm not objective you know to but my kind of view on this is like if you look at the tfu space tfu tfus space it's it's booming right now it's actually like really hitting it its cycle we're like the only people that I know of in the recreational space everyone else is in the medical space and so like my idea is like we got to get some Market as fast as possible because one like you know that's what you want to do as a company like as incentive wise but also you know if you have a big consumer device that is non-trivial right you you can develop

57:04

a war chest to kind of expand into those medical applications um so so you can use tfus um to to do slow wve um you know induction evidently sleep um actually I believe Philips had a had a device for a while um out but I don't know what happened to that um have you tried this on anyone yet so again I don't maybe I didn't mention this so this is a read only device okay we we designed this in October we had to build our model which we just showcased again like about you know two months ago the thing that's very counterintuitive of building building Hardware you know one of our advisers is uh a gentleman named Sterling chrisen he led Apple's neurotech prototyping team for the Vision Pro and it's very counterintuitive they start with the beautiful prototype you know and they

57:45

say like this is what we're [ __ ] aiming for like we need to get this to look like this and then you go and you build your your technical prototype I have pictures I can show you it looks not like that um but you know you got to do it it's cheaper and so on so forth um and then you you basically do your work of validation right of of the Three core experiences and then you merge those devices you say okay we got this we have to make it look like this how do we do that um so there'll probably be another process by which we take that technical prototype like for example I can tell you for sure the totem is going to be bigger probably be like the size of a book there's a lot more of things like a waveform generator and and and so so forth things like that um so uh so yeah more work to do for sure we are starting neuros stimulation next week I like to

58:26

be your guinea p i gu so I'm not going to lie to you 4,400 people signed up to do just that in 48 hours um but you live in New York and and so if you have a CT scin I mean I'm happy to I do I'm a practitioner So Perfect You're the perfect type of person okay getting a CT scin is a [ __ ] in New York man like I had to get we to like lie I shouldn't say that okay all right next question yeah um real first of all this is a fasinating oh sorry I can see okay fasinating subject than doing this this unforunate question which is um whatever platform you're developing that the data is going to read to as well as the device itself do you plan on either currently or eventually adding either

59:06

fail states in the device for security audits in the platform because unfortunately in my field my first thought is sure hey this is a target that has this I'm going to go [ __ ] with absolutely it's such a great that's a fantastic question so um you know there there's a few things that you can do so there's the firmware right that you you obviously you know you have also you can you know through your transducer design basically limit the threshold like deao on the hardware so that the lens it's like you're incapable of getting it below you know above let's say you know 720 um you know which is again the threshold for for uh for uh the FDA now let's say you wanted to go and do medical applications with it as well and and same so it's like higher

59:48

things again you just limit it so that it's impossible for the lens to create uh The Beam with uh within the thermal band of ultrasound um um and then in terms of other things I mean listen guys this company's N9 months old so give us some time um but you you are absolutely correct that things like security are non-trivial here um and so um there's a number of things that we will we will do obviously to figure that out um I also want to mention something that you didn't necessarily act directly ask directly but is important to say which is you know tfus on its on its own is not addictive um but experiences are right so we have good research that shows that if you have a lucid dream once or twice a week it's net positive for wrestlers so we're going to lock the

1:00:28

device so that cuz we can tell definitively with your gamma frequencies of EG um that you had a lose dream so whether it's one day or two day we'll figure it out positive mood obviously [ __ ] addicting right like I mean you have a device that can just make you elated now there are people that argue with me and say like why is that such a problem like why can't you why not just keep people like why can't why take that away from somebody um but I mean one then we're certainly going to get [ __ ] you know by the FDA they're definitely going to come in um and and two like is about a lot more than just being happy um my my grandmother from Europe always just say Americans are obsessed with you know like happiness but like like life is actually a much more multiactive thing um so so one of the things that I think I I would like to see because we

1:01:09

do need to regulate this I mean again it's the I'm like you know not very keen on regulation but this is like sufficiently you know important where it's like you got to have like you know something and I think it I'm less concerned about the threshold I actually think the 720 should be raised quite considerably um um you know very safe um but I think it needs to be regulated around the experience itself right like cuz Focus not [ __ ] worried about that right like it's better than Aderall like you know for you um so depending on that uh you didn't ask that question but just want to add that in there that's great how many more questions are there okay so we'll just keep that in mind okay go ahead yeah this might be I

1:01:53

already took like you mentioned earlier there's like a subgroup of like Brian Johnson like um you know Maxis I guess and So within that subgroup a lot of us really care about this sleep score concept I sleep I agree yeah I'm curious from a perspective of like is there some sort of trilemma or like what happens when we have more allocation of our sleep to lucid dream does that impact does that impact further like our sleep sure as I just mentioned you know there's good research that shows that lucid streaming once or twice a week is net positive for rest for this so again that's why that'll guide us where we're going to rate limit it you know in a week um we

1:02:33

also write again to give you the sequence of events when you're wearing the headband in the context of lucid dreaming is it's looking for REM you enter REM it the pot it the P it's looking for the gam frequencies it's like are you Lucid okay you're Lucid it keeps you lucid and then you exit R and then it turn turns off um there will be some lasting effective that'll sh for a little bit but the idea is obviously to allow you to just continue with your regular uh you know sleep cycle right where you continue you and then you'll have REM cycles that aren't loosing you know um so um again let the like we're going to like that's the research as of right now once or twice a week that's what's going to guide us um and so yeah like I mean I'll be honest I'm not even as worried about um restfulness as I am worried

1:03:16

about like somebody losing their mind right I mean if you drop acid every day what do you think's going to [ __ ] happen yeah um and so I I think if you had a technologically induced lose a dream every single day of the week for a month you know I don't want to meet them um I don't know that oh wait wait wait one last thing who has who has tweeted I'm going to be annoying for a second I H up LinkedIn and then no one else who did everyone else we got we have only had one tweet so far which is terrible for this people what can I tweet later no Tweet now otherwise you're not going to do it later we have to do it now I'm going to be a little bit of annoying so take out your phone and like actually do it

1:03:56

um because it'll be it helps it helps the community did you do it about to we'll do some in the B then go

1:04:17

ahead so so the definition of elcid dream is is the awareness that you're in a dream that begins to control the phenomenological consquence of that dream that will so if you having a night because you're a stus you want to have [Music] a all right in the back aam expl you speak up I'm you have a hug parameter to explore so what's what's the evidence on the region you have to keep to actually produce so again the neural correlates of so the way that you do neuros simulation regardless of the medium right is you have to understand at a minimum the neural correlates of a given brain state and moreover it needs to be

1:04:57

a discreet Universal brain State discreet as in it is one thing and not another I'll give you a counter point unfortunately flow states are not discreet right you could be a surfer and enter flow and it's your motor cortex you could be a chess player and enter flow and it's your spatial reasoning so it' be much harder one to create a model of it and then all you know to to simulate it Universal simply means right that all of our brains look the same in that experience um now you mentioned something about the parameter space which is very you important you're absolutely right so we pre-train the model right and what you can think about that is it's creating like this this Vector space of of hopefully successful potential patterns of neuros simulation sequences of neuros simulation but it's unclear right you know maybe you're in some part of that you know Vector space

1:05:38

and I'm over here so we think we can get pretty good generalizability out of the box but to make it improve over time the e in the headband does reinforcement learning so we can Zone in kind of you know uh you know kind of get into finding where you are in that broader Vector space it's very right like when you when you open Tik Tok for the first time don't do that um and I don't give [ __ ] um and and and you know it doesn't know you and then you start using it and it learns about so that's question are am that's not a really answerable question I don't think right now may I mean the you it's unclear what you would

1:06:19

like the units of vector space that you would use to describe that well in relation to how fast explor like with the device like we have evidence that with three sweeps in one night you can hit like the switch for me differently from you or oh oh I see okay well I guess this is what I say no no that's fine um okay so for example focus and positive mood like we already have the values the research has been done we're just replicating it focus and lucidity are extremely similar right focus is increased activity in the prefrontal cortex while you're awake Lucidity is the pre prefrontal cortex activity increasing during r so um we have a lot of reason to believe that in quite quick succession certainly

1:07:00

focus in positive mood um but again like the key difference is like we just need to go to sleep with it have the EG like understand the r REM and send out polies but um yeah you were talking about the gamma raay being kind of the notam notam camera weight being what tells you if you're if you've gotten some into so are you only going to be optimizing around getting that or is there going to be any other kind of input that the user gives of like because I mean do we know for sure that it's like a binary like either Lucidity not Lucidity or could it be that like one of those you know po Holograms of the ultrasound results in a certain

1:07:41

flavor of Lucidity and a different one results in a different flavor of it and so you'd want the individual to also be able to give you some more like qualitative data yes so we have a survey that when you when you end the session we use the same survey that's been that's used at universities sorry um and and so you'll give qualitative feedback that'll be paired with your your data for the reinforcement layer and that'll allow us to improve things over time we'll also do that of course prior to the launch of device um but you asked something about whether it's optimized specifically that so depending on the brain state that you're that you're looking to actually you know stimulate it'll be optimizing for different things but gamma frequency

1:08:22

in the context of lucid dreaming is the thing you're optimizing for okay all right over here behind is there anyone in the back I feel like the back is sometimes not cared about anyone okay go ahead okay I was going to ask there someone in the back there's someone in the backing is done yeah yeah yeah yeah so we do some yeah yeah yeah so the the EG processing is done on uh through this neuros Sky chip it's about $9 a chip make it bigger right yeah that's not because of the chip it's because of uh other uh things that we need uh sensors but but waveform generators other things yeah um now um

1:09:04

and then we do the reinforcement learning um and so so forth VI the cloud um the device connects by bluetooth uh and or Wi-Fi um you know through the app um you know you don't want to you the problem is is like if if any of you guys have used the VR devices the issue is that they have to do so much compute on the device that the heat dissipation right on device is really important like like a lot of the R&D for the Vision Pro and I know this you know pretty definitively is like right was like how do we create a heat dissipation you know capacity uh you know because think about how many I mean they're running like what 5 NM chips you know whatever I mean the size of the battery like literally right it has to be tethered all these things so um you really want to do the

1:09:46

compute up device if you can all right any any more questions oh there's someone in the back okay in the back go ahead did you guys tweet guys in the middle tweet tweet it actually it so helps us so much I sorry to be annoying but like if you want more people to be here next time and for it to be a better crowd for you uh you have to like you have to like keep keep the growth up otherwise it it dies unless there's pedal to the metal okay so tweet it really it makes it it makes it better for everyone I promise um yeah in the back one of the concerns of people who losey dreams like sleep paralysis is that anything you guys have worked on looking yeah

1:10:27

um you know it's something that we're going to have to work on but frankly until we get to uh induction again which we start uration next week you know that that that I honestly say is an open question um you know I've been lucid dreaming for a long time I a lot of people in the lucid dream Community sleep paralysis is a very different thing I mean right so when you're in rem your motor CeX is uh you know disabled um and a lot of people think in in in sleep that you're like awake and can't move but actually you're you're you're still asleep in rem and your motor cortex um you know prevents you from like waking you know you just can't wake yourself up I bet you that if that was a thing I wonder frankly if we could not

1:11:08

figure out a way to use tfus like should you end up there to get you out of it um you know in the sense that like you would just really like blast somebody's prental cortex like and and wake them up um or or we can trigger it with a noise right um you know or or a light um but but yeah something yet yet to be of really figured out what is so I mean the experience I I don't know i' I've earnestly not um read the research right I shouldn't speak like too definitively on it experientially right people report um where they they basically feel like they you know they're awake they can't move and they see like a a like a dark figure let's say and they're very scared but they can't like [ __ ] move and they also

1:11:50

can't wake themselves up kind of suck yeah all right awesome stuff thank you so much for making the time Eric I'm Muhammad by the way freshman at Amber Aeronautical University and I'm also working on developing a space shield with melanin so I'm curious about I'm curious about how does that help me why do I need that even as a student or you mentioned that it can help you um write a code write code um or make an app write yeah so like is it possible that I could just loose a dream and then develop something and then just come back to portion of the talk right like I mentioned like all the discoveries and

1:12:31

things like that and creative Pursuits Etc that people have done um there are certain things that are like a lot harder to do in elusive dream um reading is like [ __ ] impossible like you know the letters are kind of all over the place um you know there there are I I know I suppose there are limits I think we'll learn more also about what those limits are like what you simply cannot do um like you couldn't do material design I doubt it right I'm sorry I mean you mentioned C so I thought you might be able to do take take that problem and break it down to its most like kind of anatomical thing or something I really don't know but what what I would say is the history of dreaming and Discovery and creation is quite long um and again if you go to the locid Dre sub which has over half

1:13:12

million people in it you'll see people talking about a lot of things and really mundane I mean I hate to say this about the professions but they might have very normal professions and yet they still use their you know their their lucid dreams do and and you know we we we've talked to you know people about training for Special Forces this the other twoo language acquisition so like what's your ultimate goal with this device and how is it going to make us better humans um so I mean I mentioned like re-enchanting the world right um lucid dreamers report far less rates of depression far greater s sense of personal agency that makes a lot of sense right you like you're a God in your dreams and you wake up you're like [ __ ] this like I want I want my life to be as malleable as that um and so there's a lot of positive you know

1:13:52

effects on people um what I want to do is I I want to one understand what Consciousness is and I think that this is a method by which we could do that think about Darwin right what did Darwin do he goes on the HMS Beagle goes to the [ __ ] galpagos and like looks at a bunch of finches and he's like Evolution you know I'm over simplifying but the question I ask myself is you know who are the darwins of Consciousness like who's to say that you can't go into a lucid dream and do something very analogous to that just sit and watch um I don't know we'll see are the effects lasting or are they how long are the effect so depending on the frequency you use the ultrasound could you know be like like the effects could be like a minute or or 20 minutes um but but but to be

1:14:35

very clear when I say that like as you exit REM for example turns off um you know you're not like you'll be able to you know with the slow wave sleep somebody asked a question about like slow wave sleep that was uh slow wave uh simulation to induce uh sleep you know you can you can let it kind of still continue to have effect while you still have these Cycles right um so this might be a crazy question but so like you're collecting you listen to what I said I was crazy you C question this is so you're collecting a bunch of you can you are collecting and you can collect a bunch of more data with these these devices um have you given any thought to like using Federated learning or some

1:15:16

other AI like two techniques that can take into like take into account many different data sources and like coordinate lucid dreaming amongst many different people to have like Collective lucid dreams okay um purposeful sure let me explain this again I've been locid dreaming for 18 years I've never had a locid dream with anyone ever okay but I will explain to you how you could do that I mean maybe we'll talk about it after like the way you could do that is like very synthetic and it's not actually a thing um in terms of your other like architectural questions if my CTO is here he could probably talk to you a little bit more about it um but we made a couple of architectural changes that also increase the amount of data

1:15:57

that we can use to to train the model where we don't just require simultaneous EG anymore but but is the data stor uh like I don't know like on AWS or some [ __ ] okay locally control so yeah that was my question so you so you you collect the data the the company collects the data from from the users sure yes so yeah again something by the way that you opt into um uh but you would want to opt into it because again that's your reinforcement learning right um so you have an incentive to do so um so yeah you okay yeah do both of them stay at the same same time and then he'll answer both of them that Transformer

1:16:37

that neurolog that neurot Transformer sorry what a neurot Transformer what about it you think it could be used for or what what else you think it could be used for I mean so and then what was your question okay two sem related questions say it okay okay so I just want to make sure I understand how this works right so you're reading the brain State you're Computing a state transition that would bring it into vidity and you're using ultrasound to apply the transition is that right uh say it again you're looking at the current brain State using your EEG okay and you're Computing a transition a state transition that BR the brain into Lucidity and you're going to apply the transition using

1:17:17

ultrasound yes um yeah I think the answer is yes I I kind of I'm having an issue with the the phase transition right like you're looking first for REM uhuh then you turn then you send out the tfus then you're looking for gamma frequency so if that's what you mean by phase transition then yes okay so question two um there's some Pap release recently I think one of the researchers came and I talk about it where they use ultrasound like open up the blood brain barrier or to deliver like drugs yeah but like the unfortunate side effect of neuro inflammation cuz things are getting in I shouldn't be in there like is there any risk of that happening yeah I'm I'm very familiar with this we're we're we we we uh we're not at the frequency that will that will

1:17:58

do that I actually know the person you're talking about that yeah speaker T 11 yeah yeah yeah okay go ahead ask your question again okay what were the other uses for the or the potential so neural transfer me right like so my co-founder Wes um you know I actually found him because he was doing neural art stuff with Grimes and he was using an EG in doing neural art so like you know using EG running it through a Transformer um and uh I don't know creating some displays based on her EG signals I suppose as well as his pianist that she performed with um you can do like neural feedback therapy games um you know so on so forth again um you know this is the worst they'll ever be I I I couldn't possibly presuppose you know what

1:18:38

they'll become is that all the questions do we it all right did everyone tweet about it un you did we have we have someone who's trying to I think we're texting your friend anyways oh wait I forgot our the the the gift has lucid dreaming changed your life has lucid dreaming changed your life um yeah um absolutely um that's probably why I started the company right um it again I think the key thing that I I mentioned before is like this re-enchantment of my like the waking World um but like you know I lost my grandmother two years ago lived be 98 had a wonderful life what do you think the first thing I did was when I had you

1:19:18

know had a I want to talk to my grandmother um that's a really powerful thing um I'm grateful for that um and and and some of the most profound like spiritual experiences I've ever had have been you know so awesome ever talk about acons your device uh definitely not how long something sure so um you a r cycle 20 minutes so that's a maximum one thing Inception did get right um right is that there is an experience of time diving um so it

1:19:58

varies you discount disorders a part of the lidity score say again do you do you discount disorders as a part of like a high Lucidity score discount disord like say like take a sample of people say schizophrenia sure um definitely one of the things that you're going to have to fill out on the onboarding process is whether or not you have epilepsy this that the other like and that you're assuming the risk of that amazing well let's give our speaker a round of applause and we have a Dei NYC speaker mug for you and where's Sanu are you guys come come get a d this is is

1:20:40

Speaker mug and it has D NYC sort of painted on and we had to get these like custom made and it was not trivial thank you yes of course so uh thank you guys all for coming uh feel free to converse take us slice of pizza uh send me feedback oh the other thing is like for real like this these aren't going to happen unless we get more volunteers so uh make sure quiet quiet make sure to volunteer for the thing if you possibly can because it really helps out Mak work okay thank you guys all right should I just turn it off cut