breaque

Multiple friends have now told me a version of the following story from 1st hand experience:

1. Employer enacts an everyone-must-use-"ai"-for-everything policy
2. The bills start coming in
3. Employer goes "wait no not like that", starts severly restricting token quotas and/or telling employees to only use lying autocomplete "for important things"

..and this is with the current rock-bottom loss leader pricing, where the real bills are footed by the investors of the slop machine peddlers.

How do you think it'll go when they eventually start cranking up the rates to cover their costs, let alone make profit?

Yes, I know: the endgame is to wait for the models to get "intelligent" enough, then save money by laying off everybody except for a single intern, who the CEO can then task with prompting the company's product line, business logic and operations into existence. Unless your product is slop, good luck with that.

Mar 25 at 8:22amWeb
Show threadtotetmatt2d ago
@breaque Yup, better pay +2k/month per engineer for LLM as accountability will allow you to manipulate it to maybe reduces taxes e.g, rather than having the exact same amount as salary increase and pay social contribution :/
Show threadtotetmatt2d ago

@breaque And be prepare for the "Token Optimization Engineering" or "Framework X Y Z to reduce token Consumption".

Sort of half joke that running currently : Learn Chinese as it consume less token :)

Show threadPaul Grenfell2d ago
@breaque It's the "you must all use AI to increase efficiency, and we'll monitor your usage to make sure you're using it" thing that gets me. So your KPI is token usage? Not sales outcomes or revenue or bug counts or anything like that? It's like telling coders that they get paid by lines of code.. because that generally works well, doesn't it? 🤦‍♂️
Show threadunkx2d ago
@evilpaul_atebit @breaque the KPIs are whatever gets management their bonuses, and measuring a number that says ”Tokens!” is a lot easier than, you know, doing business.
Show threadPaul Grenfell2d ago
@unkx @breaque Exactly. And the cynical side of me wonders if this sort of KPI is forced in at companies where the investors are also invested in AI... two KPIs with one stone? DOUBLE BONUS! 
Show threadunkx2d ago
@evilpaul_atebit @breaque I’m fairly certain that’s one reason. Though most likely it’s more often just greed, stupidity and fomo. And also because the AI told the CEO that that’s an excellent idea, and that honestly, they’re not just leading, they’re a trailblazer actively innovating in the complex, dynamic and exciting digital frontier, and that they should be proud of the fact that they are the smartest CEO to have ever lived.