Pylimitics

Nikhil Suresh wrote a thought-provoking essay that introduced a term to me (I don’t think it’s a brand-new term) that really resonates: AI mania. Suresh is a director at a software engineering consulting firm. I think they’re pretty well connected, such that he’s often in conversation with top executives and directors at major companies we’ve all heard of. 

The essay is AI Mania is Eviscerating Global Decision-Making. It’s a great essay, notwithstanding the title. The gist of it is that executives are, en masse, touting the effectiveness of AI in their companies, not because it’s true, but because the social networks they move in demand that they say so, at the pain of excommunication and banishment (and relative impoverishment). Suresh adds that the executives he’s spoken to privately, and even considers friends and acquaintances, often really believe that AI is living up to the hype, even though they have (or should have) plenty of hard evidence that it just doesn’t. 

Part of their belief comes from the social effects further down the corporate ladder. When I worked in a major tech company, we — and I mean every single employee, regardless of role — were strongly encouraged to use AI. They gave us unlimited-token accounts. They bought their own AI companies and built an internal AI assistant to “answer internal questions” any employees might have about where to find something, how to do something, or who to ask about something. There were a lot of questions like that, because this was and is an enormous organization. 

I’ve since seen official and semi-official pronouncements from that company, on LinkedIn and elsewhere, exclaiming about the wisdom of building that internal AI tool, and how well it works. I was there until early this summer, I used the thing, and here is my in-person report: it. does. not. work. I never got a single useful response from it. The existing “search the intranet” tool was also pretty bad, but was far more effective than Alvin the AI (not its real name, sadly). 

In his essay, Suresh identifies a compounding effect I never thought about: if you’re a huge vendor of products and services, and your customer executives are touting AI in their organizations, you can’t afford to alienate them by disagreeing. It’s all one fairly tight social circle, and you all stick together and agree.

You see this kind of behavior all the time in big corporations; the business fads, the new engineering approaches, the newly mandated software tools that replace the previously mandated software tools without being better or even particularly different. Suresh calls it “cultish,” which is pretty strong, but I can’t disagree. You especially see it on LinkedIn, the group home for experts newly excommunicated from the big (corporate) house. Suresh puts it this way:

“…these people [executives] have exhibited a pattern of behavior…crafting management environments that I can only describe as cultish, ineffective, and ‘please dear God, do not let it be on earth as it is on LinkedIn’.”

Part of the message here is that too much of the bloviation about AI is just hype. The other part is about the job of executive teams, at least in prominent tech corporations. In an organization of a certain size, the purposes of the jobs at the top are primarily social and entertainment-related. You show up on the business entertainment shows, like Jim Cramer’s Mad Money. You deliver speeches and presentations. You rub elbows with folks with similar roles in other organizations — both commercial and governmental. If you pick the right people to watch in an organization, you can see them catch on and adapt to the role they want. They learn the speech patterns, the vocabulary, and they get better at the “being an entertainer” part (I don’t mean to denigrate this; after a long corporate career I can admit that it can be fun and gratifying when your executives are good entertainers. It’s important.)

The disquieting part, though, is the cultish behavior. “It is difficult to get a man to understand something when his salary depends on his not understanding it.” (Upton Sinclair, 1934) When the salary is astronomical, as executive pay has become, it seems like the difficulty might be multiplied.