The companies laying off staff for AI today will regret it in five years | Hacker News Reader

The companies laying off staff for AI today will regret it in five years

fortune.com

11 pointsekpyrotic8 comments

The whole premise is a category mistake, companies cannot regret things. And even if they could, it's irrelevant, it can absolutely be rational for today's executives to take a course of action that would likely need to be changed in a few years.

EDIT: typo

It's rational for executives to take a course of action that will maximise their incentives - which are always short-term for some reason.

Isn't the reason their compensation package? Maybe if they had to buy certain number of stock and lock it up for some reasonable period say 10 or 20 years things would take longer term view.

You’d have to change comp across the board. No one is going to want some draconian lock-in package when the other guy across the street makes a king’s ransom in a quarter.

“Ratchet, ratchet, and bingo,” as they say.

The company persists. The incurring of future cost and the opportunity cost of not having those junior staff will tell on future value.

Sure, the current board and c suites don't care. Your point is true. But the company is affected and it will be told as it goes in and out of the market and private equity: otherwise valuable, it will be less valuable on this score.

Larry Ellison won't care, but not all companies are Oracle.

Many companies are not publicly traded and have a single owner that is interested in the success of the company over the span of decades

I agree, just like when companies decided to go 100% outsourcing. The first few years are ok as company knowledge still exist in the company and in the external provider. At around 5 years, thats when enough of the company knowledge and culture is lost.

AI will most likely face the same, unless they can come up with ways to integrate the AI into the business. Similar to how people would benefit hugely from having personalized AIs that look out for them and their families. It would need to be more then just a "theme" on top of OpenAI / Claude etc; It would need to run locally... somehow. A lot of businesses can afford to buy or rent the hardware required, smaller businesses would probably need to use someone else's service for it.

What we call AI is still in its infancy and its going to be interesting to see if we get another AI winter and how much of the current tech sticks around. The LLMs we have today can be very useful if used responsibly, but we need something more around them to make them useful personalized AIs that can run on.. something at home and in the businesses datacenter.

Few, if any, of those making the AI decisions will be present in 5 years. Many of the companies will not be present, either. A 3, maybe 6 month window is like “forever” for such people. So, given this reality, catch the wave and sell (right) before the crest.

Reply on news.ycombinator.com