← All postsBlog<br>Accountability and AI<br>Reagan McCullough·August 14, 2026

A stalwart topic in the entire AI discussion has to do with the number of jobs this new technology will take. Elon Musk has claimed that the scale of job destruction will be so large as to REQUIRE a universal basic income. No one can deny that Musk's companies' involvement in this space gives him a unique, "under the hood" understanding of the technology. Citrini Research published an article outlining a thesis of the job destruction being so vast as to cause massive demand destruction resulting in serious deflation. Yes, deflation. "Sensationalized" strikes me as a very fitting adjective of the claims that are being proposed by the aforementioned parties.

Before the wrath of the tech world comes bearing down upon me, allow me to clarify my position. AI already has and will continue to reshape the landscape of work. Clerical and administrative tasks will certainly see a reduction, if not an outright elimination. Other jobs will simply see a restructuring of what the employee does versus what the machine does. We have already observed these changes, with the day-to-day tasks of certain roles being flipped completely on their head.

What must be noted is what the positive and negative incentives of an individual's job are and how they affect the person in that job's approach to their work. You as an employee have the positive incentive of doing your job well. This roughly parallels the way AI models today are constructed. You as an employee also have obligations outside of work such as rent, car payments, groceries, utilities, personal self-worth, prestige, and a multitude of other motivations to get you out of bed in the morning. Claude and ChatGPT don't have these. Botching one task can cost you as a human your job, and with it would go your ability to pay for a roof over your head, a car to get around, food to eat, and your self-esteem. What is the skin in the game that an LLM has? Effectively none.

Nobody seems to consider the downside of your AI agent paying the wrong vendor a huge sum of money. Nor do many consider the possibility of it improperly architecting an internal software program that could cause huge inefficiencies in how a company's workflows run and take a ton of man hours to untangle and rebuild. Either of these flaws would give a company plenty of reason to show the responsible employee the door. On the contrary, the benefit of the doubt is the favored approach with the faceless, childless, and amorphous computer algorithm. These are all possible scenarios that throw a wrench in the forecasts of mass-scale layoff forecasts.

Can AI keep being improved? Yes. Will it continue to be refined? Certainly. Can it ever possess the traits and have the conditions that give humans incentives, both positive and negative? That is doubtful. Skin in the game and personal incentives are factors that the wise man heavily weighs, and the fool never notices. This is why technologies like AI, despite their capabilities, will not usher in the visions of society that Musk or Citrini have.

Get started<br>Try Authoryze free<br>Connect your AI agent in minutes. Rules enforced server-side, cards issued per transaction, human approval for anything over your threshold.<br>Create free accountRead the docs

Accountability and AI | Authoryze Blog