Education and redundancy are well-known and documented elements of risk management. So well known, in fact, that ISO 31000 has included them as points in the standard document, meaning experts in the industry also consider this a vital part of the system working.

The sudden adoption of AI agents and the promises of pushing more work onto senior team members break this system by trading short-term gains while sacrificing the path to the creation of institutional knowledge. 

In reality, the more apparently capable our AI systems look (yes, apparently capable), the more knowledge we require to control the output being generated by the machine. Which means that as seniors start taking pensions or simply leaving the company—an event that is ridiculously common—we have no built-in security to replace what this person was doing before.

Another problem might be that: the more Western populations decline both in age and production capacity, the more visible this gap will become, as entire industry and production sectors weaken at an incremental rate and have little opportunity to recover. 

Sure, investor numbers are padded now, but within a decade this will compound with the unrealized gains of extreme AI implementation in products that not only didn’t need AI but are dependent on the APIs of others—as most things in the LLM space are.

Yes, you can sell an agent today, as long as your capacity to prompt is not of interest to the major players in the industry. And if you think this sounds unrealistic, try taking a look at the interactions with MS Copilot. All of them say “powered by GPT–current version.” Capitalist dystopias are certainly not something I believe I will see in my lifetime, but in reality today’s corporate empires aren’t as stable or independent as we might want.

On this particular occasion, the fault lies with the shortsighted policy of cost-cutting tied to personnel reduction at the lower levels, where juniors are educated, under the pretext of automation. This is, by different measures, a perfect storm with all sorts of problems.

We have now started to pretend that the intellectual capital gathered by corporations can be sustained by hope; so much, that it seems more a religious exercise than a technical enterprise; the sustainability of intellectual capital is at stake when there is absolutely no interest in its renewal, while simultaneously subsidizing the gap and refusing to provide the economic incentives to do look for or become the replacement (no one will go into a poorly paid job).

Ironically enough, as the market for seniors dries up, companies will start plugging the opening with cash, burning out their coffers and accepting ever-expanding pressure on their war chests and their bottom line, until new competition pops up explicitly to take advantage of the opening (usually led by one of the minds hired to pave the previous system). This might continue until we have a sort of gig intellectual economy, which is good for diversity but undermines the capacity of the sector to counter the effects of more consolidated, usually government-sponsored, international actors.

If you were in an army, you would never believe anyone coming at you saying, “everyone in the room is a general,” nor would you expect this same general to run field exercises himself, maintain the equipment, and drive the cars. It would be equally stupid to say that he is doing it because he now has the time for it. 

Why, then, is a senior taking on junior tasks worth destroying the whole system?

---

The Pocket AI Guide is out!

📙 Amazon US: https://a.co/d/gCHHDax
📗 In Europe Amazon Germany: https://amzn.eu/d/3cmlIqa
(Available in other stores Amazon stores too in Europe)

Check the free resources in this website!