It's not the case at all - years ago we tested Pong with human neurons vs mouse neurons, and human neurons performed better. More importantly, they allow the study of the effect of external interventions (such as trial medicines) on the learned function of human neurons. So we spent years building a machine and software platform to facilitate this sort of research.
The Doom project is mostly a bit of fun that demonstrates how quickly someone new to the field can see real results with our platform. Our data scientists believe that biological learning was demonstrated in this case.
Cortical Labs CTO here. My focus is on the system itself rather than applications, but for what it's worth ..
When the neurons didn't get stimulated by the application, performance did not improve. The only explanation our data science people has is that the neurons began to learn and perform the desired (highly abstracted) task of 'playing Doom'. This was not a surprise as we've shown this before with a version of Pong using a different platform. We built the CL1 and the CL API to enable rapid iteration on this sort of work.
One benefit to this is that when you have a measurable learning effect, you can measure this before and after exposure to an experimental drug or other molecule. It becomes possible to test the impact on neuron function, not just survival.
It was a great video. You're doing interesting work.
I've also seen implementations of realistic neurons, spiking models, etc. In software implementations, what combo of libraries and hardware would equal your 200,000 biological neurons in performance (esp training)? How many GPU's are we talking about?
(Note: If you haven't already, it might be helpful to publish a stack like that so people can experiment with encodings or reinforcement methods at no cost to you.)
We focus on using real neurons, I'm not aware of a software based equivalent. But users can `pip install cl-sdk` to get started with our API. The SDK is still early but supports playing back a recording of real data so applications can be built with a realistic spike frequency. (We'll be releasing a set of recordings for this)
For academic papers, the phrases they use include "biologically plausible," "Hebbian learning," and "backpropagation free." Searching for those with "neural networks" or "neurons" will turn up cutting-edge techniques.
We'll likely leave the answering of your equivalence question to our users. While we do plenty of internal research, the primary goal of our cloud platform is to dramatically expand access to this field. There are no doubt many low-hanging fruits to be discovered.
I work on the software stack for a biological computing platform (think: tool for programming with human neurons in a dish).
Coding work spans FPGA (SystemVerilog), Linux kernel C, userspace C, Python, and yes, some web services and Browser JavaScript also. I also work on the network engineering of the cloud service and on the Linux OS image.
Easily the most fun I’ve had as a developer and I’ve worked on lots of different types of commercial software projects before. Not all the world is web apps, embedded work can be very satisfying if you’ve not considered it.
The neurons live on an electrode array kept inside a machine which keeps them alive (body temperature, sugar, gas, etc). The embedded system continually measures the voltage at each electrode, and can electrically stimulate the neurons via the same.
Users are provided with an easy to use Python API that allows realtime observation of the raw signals and of detected action potentials (spikes). They can also execute precisely controlled stimulation sequences, which allows information to be presented to the neurons. Then there are a lot of higher level application APIs to handle things like recording data, real-time visualisation, and experiment config management.
Why: it appears to be possible to teach neurons to perform tasks (we taught some to play pong several years ago). Since then our focus has been on building a system that enables rapid iteration in this sort of research. We're selling the system to organisations that have the appropriate biology skills, and renting access to hosted devices where we provide and take care of the cells.
Woooah. Impressive. Sounds like a very fulfilling and interesting job! Love anything that makes software integrate better or impact directly the physical world
If a programmer is more efficient with AI then you need fewer programmers, assuming a fixed amount of work is needed. So in that sense AI would be replacing programmers.
This reminds me of the time when Quake started rendering inside the start button of the Windows 95 desktop (or maybe Win 98). I wish I could remember the details but it think it was something to do with alt-tab.
That quirky effect is a known quirk of early DirectDraw/WinQuake on Windows 95/98. When a game grabbed the primary surface in certain modes, parts of the desktop GDI (like the Start button/taskbar) could briefly be overdrawn or show the game’s backbuffer due to how DirectDraw managed the primary surface and cooperative levels in fullscreen and windowed modes on those systems.
This sounds incredible and makes me want to go through a similar learning process. I don’t suppose anyone could recommend a book or course along these lines?
A lot of us started out putting together some of the circuits in Forrest Mims Radio Shack books [1]. He does a good job of explaining how they work as well.
And then there is an amazing site with (thousands of ?) electronics hobbyist magazines archived [2]. If you want to start browsing, they often ran series in their issues along the lines of "getting started in electronics". Regardless, some awesome projects in there.
But if you want to go back pre-transistor (ha ha) the U.S. Navy had a great series on learning electronics [3]. Def. analog.
The Art of Electronics is an incredible resource. They have a companion book which guides you through hands-on labs. I have not read it, but I trust that it would be worthwhile.
I've never liked the idea of community actions in the critical build path, so I use official actions/* when I can, and otherwise use actions/github-script to invoke the GitHub API via inline JavaScript when I can't.
> This is in spite of having a long list of technical accomplishments and an impeccable public track record under my belt.
If skills aren’t the problem, one possibility is that a rigid attitude, lack of humility, or something like that is rubbing interviewers the wrong way. Please forgive my unsolicited advice and good luck with the search.
Or honestly maybe just bad luck. Lots of possibilities for just a one-off internet comment.
But yeah no doubt wise to do some self reflection and analyze what one could do better when trying again. One shouldn't just continue the same strategy without reflection when faluire occurs. But also don't be too hard on yourself, sometimes things just don't work out.
I hope it's bad luck because I struggle to find what else to improve. I can launch high quality projects quickly and reliably and my main concern is product-market fit. I feel like I've tried everything and its opposite. Problem I face now is I've lost a lot of faith in the system and it requires a lot of energy to maintain my work ethic and also maintain the outward appearance that I still believe in success. I often feel like I'm only role-playing entrepreneurship and there is no chance of success.
Sometimes I feel as though I'm being hindered artificially by some hidden forces but I try not to focus on this. It's partly why I share my experiences online and why I talk about this with my co-founder and family members. It helps when others validate my experiences because it can often feel like too many coincidences.
I worry the feeling might get worse with AI now literally manipulating people's psychology and thus, the markets as well. There's a chance my fears may become reality.
It'd be silly to keep working on a startup if there was a super-intelligence preventing me (or any new industry entrant) from succeeding right? This is a new reality I may need to factor in at some point. Would have been insane to think that way 5 years ago. Not so insane anymore.
Sounds like you might be burnt out and could use a break. If you're just roleplaying your belief in success at this point, maybe it shows in your job hunt and is the thing getting in your way.
That said it's tough out there and luck/timing is a huge factor. Be kind to yourself
