Artificial intelligence #
In the 60s/70s, in the first wave of research on AI, the field was mostly concerned with more philosophical questions of machine consciousness and the ability to reproduce, simulate or approximate the human brain in software. Lately it’s come to refer more to machine learning, in particular computer vision and NLP, and in the latest intense wave of hype around the subject, LLMs.
As time goes on I get more and more pessimistic that something good is going to come out of this current AI hype bubble. Artists and creators everywhere seem unified in their terror and anger about it all, about this deal we have apparently all collectively made that it’s OK to just feed everything we have ever created collectively as human beings into some algorithm, and then profit off its output. But even absent this huge moral problem, the whole thing just seems to inevitably lead to further and further “bullshitification” of the world: everything becoming cheaper, crappier and more and more full of bullshit, lacking a human touch, thought, empathy and wisdom.
This article, maybe a little less bullish even, more or less sums up my thoughts on AI at this time.
Links and resources #
- A nice post explaining how the GPT large language model works.
- A pretty crazy post on using ChatGPT to generate a reducer for Redux in React.
- Nick Cave reacting to a ChatGPT-written song penned in his style.
- A bit of insight from Hackernews posts into the potential differences between the blockchain bubble and the current round of AI hype.
- A nice explanation about why most of the current AI startups have scarily thin moats.
- A nice benchmark for whether AI can solve random Github issues.
- A nice walkthrough of how ChatGPT can help you write code (I’m still not totally convinced).