Hey chief I saw @Andy Beach tag this in so was happy to find your stacks..
I’d gently like to contribute a bunch of stuff to this. I did AI at university when the subject was 50 years old (early 90s) and in reality not much has happened to AI in the time since.
Transformers (2017) were genuinely novel, but most of what’s happened since is scaling - more parameters, more data, more compute. That’s not the same as fundamental advances in how machines think or understand.
What you are really chronicling is the US history of the GPU.
to make it a history of AI, where’s:
• Turing and Colossus (UK, 1940s)
• Cybernetics and Norbert Wiener
• Ivakhnenko gets a brief mention but the Soviet/Eastern European contributions are glossed over
• British computing generally
The article conflates computational throughput with intelligence advances. Each era is framed as progress toward current LLMs, but it’s really just: “We got GPUs, then TPUs, then bigger clusters, now we have ChatGPT!”
This is why I think the modern phenomenon should be determined as EI (extended) rather than AI which would better be described as ‘autonomous intelligence’ and we are absolutely nowhere nearer that than we were 80years ago IMHO..
Brilliant. What if Swift's "thinking machine" had more direct infleunce on early engineers? We might've seen actual AI concepts emerge even faster than the ABC's foundational work.
A fascinating area of potential discussion regarding the influence of what is today considered Arts versus Sciences. This segmentation may arguably be a side-effect of the philosophical schism between mind and body that emerged with Descartes. Many of the most prolific scholars of history were so accomplished, I would suggest, because of the interplay of both Art and Science. I cannot comment on the training of historical engineers as much, but modern schools have a strong segmentation between the traditional Arts and even Sciences versus the Engineering schools themselves. So perhaps, as you suggest, if the interplay of literature and engineering were more pronounced progress - including ethics and societal factors - might have co-developed.
Hey chief I saw @Andy Beach tag this in so was happy to find your stacks..
I’d gently like to contribute a bunch of stuff to this. I did AI at university when the subject was 50 years old (early 90s) and in reality not much has happened to AI in the time since.
Transformers (2017) were genuinely novel, but most of what’s happened since is scaling - more parameters, more data, more compute. That’s not the same as fundamental advances in how machines think or understand.
What you are really chronicling is the US history of the GPU.
to make it a history of AI, where’s:
• Turing and Colossus (UK, 1940s)
• Cybernetics and Norbert Wiener
• Ivakhnenko gets a brief mention but the Soviet/Eastern European contributions are glossed over
• British computing generally
The article conflates computational throughput with intelligence advances. Each era is framed as progress toward current LLMs, but it’s really just: “We got GPUs, then TPUs, then bigger clusters, now we have ChatGPT!”
This is why I think the modern phenomenon should be determined as EI (extended) rather than AI which would better be described as ‘autonomous intelligence’ and we are absolutely nowhere nearer that than we were 80years ago IMHO..
Thanks Dom for the comment and participation. Welcome!
Brilliant. What if Swift's "thinking machine" had more direct infleunce on early engineers? We might've seen actual AI concepts emerge even faster than the ABC's foundational work.
A fascinating area of potential discussion regarding the influence of what is today considered Arts versus Sciences. This segmentation may arguably be a side-effect of the philosophical schism between mind and body that emerged with Descartes. Many of the most prolific scholars of history were so accomplished, I would suggest, because of the interplay of both Art and Science. I cannot comment on the training of historical engineers as much, but modern schools have a strong segmentation between the traditional Arts and even Sciences versus the Engineering schools themselves. So perhaps, as you suggest, if the interplay of literature and engineering were more pronounced progress - including ethics and societal factors - might have co-developed.