It recently occurred to me that I might be able to teach GPT4 to add numbers of unbounded length just by describing ripple-carry addition (in context). I wanted this to be a test of GPT4 s intelligence rather than a test Continue reading
Self-guidance, or simply guidance, is fundamental to systems for image generation including both diffusion-based image generators, such as the well known DALLE generator, and autoregressive image generators such as CM3leon. A technical definition of guidance is given below. While Guidance Continue reading
In a 2014 post I argued for the servant mission as a solution to AI safety. The basic idea is that there are many AI agents each one of which has the mission of serving a particular person. I gave Continue reading
Wittgentsein famously stated that the meaning is the use . Unfortunately no analytical explanation of this statement is provided in the posthumously published Philosophical Investigations. Wittgenstein s point seems to be that language is not appropriately approached through logical or analytical means, Continue reading
Behaviorist psychologists refused to talk about mental states or internal computations in human brains on the grounds that everything should be understood in terms of stimulus-response relationships. On the other hand linguists and cognitive psychologists assume the existence of internal Continue reading
For a couple years now Michael Douglas and I have been contemplating a meaning representation simple sentence graphs based on breaking sentences down into simple sentence components. Over the last week or so we have been playing with Continue reading
A recent NOEMA essay by Jacob Browning and Yann LeCun put forward the proposition that “an artificial intelligence system trained on words and sentences alone will never approximate human understanding”. I will refer to this claim as the grounding hypothesis Continue reading
This spring I had a disagreement with an old friend about AGI. They claimed there was essentially no chance of AGI arriving in the next, say, fifty years. I have always said we just don t know. They also wanted a Continue reading
As in previous years, teaching my course on the fundamentals of deep learning has inspired some blog posts. This year I realized that VAEs are non-parametrically consistent as models of the observed data even when the encoder is held fixed Continue reading
While teaching reinforcement learning I kept asking myself what AlphaZero teaches us about RL. That question has lead to this post. This post generalizes AlphaZero to a larger class of RL algorithms by reinterpreting AlphaZero s policy network as a belief Continue reading