'AI Mania Is Eviscerating Global Decision-Making'
daringfireball.net
daringfireball.net
#1: George Soros' concept of reflexivity, where human biases begin informing, distorting, and supporting asset prices not because of their underlying fundamentals, but because of the human biases that have contributed to their prior appreciation. As per the essay being cited, if you are a CEO committing to AI as a strategy, you will also commit resources to double down on the technology. Your own identity becomes tied to it, whether you realize it or not, and you'll keep pushing for it and maybe even ignore facts that challenge the success of your investment.
#2: Marshall McLuhan (of "The Medium is the Message" fame) argues that we need to understand communication and entertainment technologies in terms of the structure they impose on us. While social media is seen as a societal ill by many, its original idea of connecting people is fundamentally, well, social... GenAI is very much a non-social (i.e., you experience it on your own) convenience technology. It gives you answers, it writes code, and it implies an authoritative perspective that is always available to you, as an imperfect human. What will this mean for our own identities as human beings?
I am very much a supporter of foundation models, LLMs, AI, etc. but can't help and think about some of the ideas above. Curious what others think.
Is there something someone could hypothetically tell you about how AI/LLMs as a technology might come to be understood as a net negative that would sway your support?
For full disclosure, I sit on the other side of this fence, with a constant concern of the issues that we will see arise. As an exercise, I try to imagine how these issues will be outweighed by the positives, though it's a challenge tbh.
Your link is the original post by Nikhil.
This page is Gruber's commentary on Nikhil's original post.
We make sure to click the link and read the article before posting here. Please follow that rule in future.
Looking at something like Anthropic's attempt to write a C compiler in rust. That definitely wasn’t 100x. It took 2 weeks to build something that wasn’t a viable c compiler. And that was best possible scenario. A very comprehensive spec, an existing implementation to act as an oracle, and existing compilers in the training data.
Your edit is even more peculiar. Writing a C compiler in Rust? Come on. I'm obviously talking about normal grunt work. And what are you even using as a benchmark? It's not 100x compared to what?
So it can’t 100x writing an OS even though it has tons of OS code in the training set. It can’t 100x a compiler. Based on personal experience (and number of games released on Steam compared to last year) it can’t 100x game development.
At my very large tech company it can’t 100x development of any of our customer facing products (and we still have a huge backlog). Judging by the fact that the vast majority of us still have jobs and big companies are still paying for Jira, Slack, and Teams, it doesn’t seem like it can 100x anything useful.
Are you making 100x more profit or at least revenue? 50x more? 20x more?
Show me the company making even 10x more and I might be amenable to the argument that it isn’t controversial.