Software brain: manton.org

@manton I think one of Nilay’s points here is that even though many people use ChatGPT, there must be some overlap of people who use it and still actively dislike the product or actively dislike using it.

> That doesn’t mean that everyone hates it. Nearly a billion people use ChatGPT every week.

@groomsy I’m sure there’s some overlap. I think the people who dislike AI the most don’t use it, though, at least not every week.

@manton When I am writing something, a big part of figuring out what I want to say comes from the act of writing. LLMs do not help me with any of that, so I don’t use them.

If I want to uncover patterns in large amounts of data, I use machine learning tools that help reveal the patterns for me. Again, LLMs mostly don’t help with that.

@manton What if the Tech Bros thought they were harnessing software developers to build the Next Big Thing and all they got out of it was a tool that’s great for software developers? Have devs pwned the finance industry?

@manton thanks for sharing this. I haven’t followed The Verge for a while, but I might pay a bit more attention now - this is a good opinion piece.

For what it’s worth, I think MB does AI “correctly” more or less, with the idea of having a master on/off switch for all AI being the right thing. This is an interesting approach to compare with Kagi, who in my opinion also does the right thing, but offers more AI options.

It’s amazing how giving users the choice rather or not they want to use an AI helps. The completel opposite from most companies, especially Microsoft and CoPilot.

@fgtech It’s an interesting thought! I do think this will be widely useful, but there have gotta be limits… We are currently in the “throw everything against the wall and see what sticks” phase of AI, for better or worse.