Many people dislike AI
AI is unpopular. This is not understood well enough in the software world. (By contrast, employees at big tech companies, for example, tend to be very well aware that a lot of people don't like big tech companies.)
This article by Nilay Patel has become a locus for discussion of this. So, here is the excellent John Gruber discussing the subject.
I'd factor Patel's article into two main claims:
- AI is really unpopular;
- This is because AI makes the world more software-y, which you only like if you have "software brain," which is not normal.
I'd say that the first of these is correct (see Patel's evidence) but the second is incorrect, for a few reasons:
- Ordinary citizens are more "software-brained"--that is, disposed and willing to see the world in terms of databases and algorithms--than Patel says. Dating apps are completely mainstream.1 People complain about them, but their specific complaint tends not to be that a pool of potential partners is being represented as software-apt data. Plenty of people use Goodreads and Letterboxd without hesitating much about represent aesthetic objects and judgments as data.2 It seems to me, then, that even in important, personal, and less-software-shaped parts of life, people will quite readily tolerate a software-ish view of the world, if that view is reasonable and beneficial
- I doubt that ordinary citizens process AI as something that turns non-software processes into software processes. Consider a homework essay. These are already highly "software-ized," in that students think of them (in part) as files associated with grades and other software systems. Yes, if someone creates or grades an essay with AI, there's a sense in which a non-software process has been replaced by a software process. Mostly, though, I'd say that it's a replacement of something authentic with something inauthentic, and I'd bet that's the dominant way it's being registered. For many of us, those inauthentic experiences feel very bad.
- AI efficiency gains are generally not salient as such to ordinary citizens. Power users and programmers are having lots of AI experiences that are obviously good. For most others, erratic, mediocre, and (again) inauthentic experiences are much more salient.
So, I admire Nilay Patel's work greatly, and I'm grateful he and others have opened my eyes to this phenomenon, but I think he has the diagnosis wrong.
Here as elsewhere, this means something closer to "in America and in other cultures I'm connected to."↩
I can imagine, and have considered, many possible objections here. Maybe people really do have the anti-software-brain position, but express it using a different conceptual scheme. Maybe the relatively low adoption of Letterboxd-type apps is evidence of non-software-brain. I've thought about it carefully and think that the best evidence for what I see is what I say here, but it would take a lot longer to justify it more fully.↩