While I have been stunned by the emerging capability of AI (see here), the last few weeks have resulted in a series of moments that have stopped me in my tracks.
I’ll share more about that soon.
But in the mean time, the following links have helped my journey. There are so many more things I have read and watched, together with my experiences of AI in the past few weeks in particular. But these are a useful way for me to show others why this is a tipping point in time.
I have talked of this moment as being equivalent to the industrial revolution, the computing revolution, the PC revolution and the Internet the revolution. Geoffrey Hinton (“The Godfather of AI”) said that perhaps it is equivalent to the invention of the wheel. Maybe that’s right.
The actual current reality – https://youtu.be/880TBXMuzmk
Here is Geoffrey Hinton explaining neural networking and machine learning in a short, simple and elegant way – https://youtu.be/N1TEjTeQeg0
What a Large Language Model (LLM) is and how it works – a really simple explanation of the technology behind GPTs from Andrej Karpathy – https://youtu.be/zjkBMFhNj_g
The AlphaGo movie – it shows AI a while back, and it is still stunning – it’s great watching! https://youtu.be/WXuK6gekU1Y
Geoffrey Hinton clarifies that much of what we are seeing now was discovered in the 1980s. The difference now is truly large data sets and massive computing power. He believes the ChatGPT moment is one of AI entering into the public consciousness.
Here he is – https://youtu.be/qpoRO378qRY
And here is a bit more – https://youtu.be/qrvK_KuIeJk
Then, what’s coming – a really excellent overview of future of AI and intelligence – https://youtu.be/-Mca6eN81Is
i believe AI agents are going to be a cornerstone of the revolution – Sam Altman and a few others have said as much. The first practical realisation of that is Devin – a programming AI suite that uses agents to automate software development. It is jaw dropping.
You can read about Devin here and watch David Ondrej talked about it here – https://youtu.be/ZkcrLOg6lL4
David’s excitement is contagious – and his amazement is mine.
Then a long, very recent interview with Sam Altman, founder of OpenAI – https://youtu.be/jvqFAi7vkBc
Then, there is an unbelievable amount of rubbish about AI online. Matt Wolfe is very good and highlights lots of good tools https://www.youtube.com/@mreflow
Here is Matt’s library of AI tools – https://www.futuretools.io/
Matthew Berman is a new favourite – he is down-to-earth and no-nonsense knowledgeable. He does generous demo videos and he is super up-to-date – https://www.youtube.com/@matthew_berman
Then, if you want to appreciate the true brilliance of the people behind emerging AI, watch this interview by Lex Fridman with Ilya Sutskever, co-founder of OpenAI – https://youtu.be/13CZPWmke6A
There are a few people I watch and I think I can see undisputed brilliance. Ilya Sutskever is one.
And then – if you are feeling like its all happening now and AGI arrives with ChatGPT 5, Yann LeCun believes AGI is a flawed terminology and then there is a long path to “AGI” – not an event – https://youtu.be/5t1vTLU7s40
TLDR? LLMs represent language intelligence – not a general intelligence. We are a long way from human like intelligence in some generally abstracted, efficient model (the brain runs on about 25W compared to an Nvidia GPU – 500W to 1KW).
Mistral AI is also taking a different route. They follow an open-source/open-weight model route and believe in a system of light-weight, low-latency more specialized models rather than on very large model that approaches AGI. The result is that their Mistral 7B model is small and low-latency and can run on a smartphone – https://youtu.be/5fNtOsu3YvQ A big benefit of distributed models such as Mistral and LLaMA is that they are private by design.