Amr Ojjeh's Blog

Contact: blog@amrojjeh.me

A Few Thoughts on AI

2026-10-21

Point 1: Someone makes claims about a certain discipline or disciplines.

Point 2: Those claims are either to be trusted or to be doubted.

Point 3: Regardless of the brilliance of that person, humans will never unanimously trust him or her.

Point 4: Those who doubt him or her will study to prove their doubt, or to recant it.

Point 5: Those who trusted the doubters will want their judgement rather than the initial claimant.

Point 6: If that someone is an AI model, points 2-5 continue to apply.

It seems to me that knowledge-wise, human understanding will never stop being important, for the sake both the individual, who may occasionally (or persistently) doubt the model, and the faction which does the same. Treating the AI as a single authority is akin to creating a new religion, and when has the world united on that front? Certainly won’t be tomorrow, and in fact, in trying to monetize the model, the AI will likely avoid offense at all costs, which may cost it trust. And if there is an offensive model, that would also cost it trust to some people. You can’t win them all!

Many people might lose their sharpness as they rely more on AI. Some people, admittedly fewer, will avoid AI to stay sharp. Arithmetic done on calculators, after all, is not the same as making a judgement call based on interpretation and evidence. There’s almost always room to doubt in the latter regard. Therefore, knowledge of the discipline and its relevant sources remain essential.

As for the job market, AI companies here have been particularly rude and arrogant, as they’ve been actively working to situate themselves as reliable authorities, dismissing the doubters, and attempting to forcibly shove themselves in every person’s bubble, all while stealing data without due credit or compensation.

Regardless, most jobs require judgement, and these can either be trusted or doubted. Even to claim that they can be trusted requires someone knowledgeable in the discipline, and to claim when they’ve no longer become trustworthy also requires someone within that discipline.

It is possible, at the same time, that the majority of corporations will blindly trust an AI model over a knowledgeable human in most jobs anyways. I find it unlikely, but in that case, good luck. Perhaps we can all stop farming for clicks and data, and go back to farming crops.