AI Agents are actually replacing people #
Amazon reportedly is cutting (aka laying off) 30,000 corporate jobs. (article )
What’s the most disturbing part of the news?
It contains reference to a memo from CEO Andy Jassy, in which he stated “as Amazon continues to implement more AI agents, fewer corporate jobs will be needed.”
AI may be in a bubble in regards to being a chat-buddy, or solving the mysteries of the universe. But when it comes to Agentic AI and bridging the gap between LLM and practical application, I’m not so sure. There IS a large amount of value in agentic AI today.
Will AI pay for itself? #
It is easy to see how Amazon can justify their AI expense by using agents to cut headcount. If you are paying $100k/year for an HR resource to do a job that can be done by an agent, then using an agent results in more profit.
However, its a knee jerk reaction for many to assume these layoffs are for the simple purpose of lining Jeff Bezos’ pockets, when the reality of the situation is that Amazon realizes if they don’t start doing operating like this, they will lose out to a competitor who does.
If they don’t, someone else will #
And that is the dangerous state of AI today, both in it’s development and it’s implementation in the workforce. In both cases the drive forward without deep consideration for the consequences is driven by the idea that your competitors will also be driving forward, and you will lose advantage if you don’t join them.
Today’s state reminds me of an old zero-sum game theory called The Prisoner’s Dilemma in which two parties are separated are presented with a choice to turn on their partner, or not. The way the game is structured, you are both best off if you don’t turn on each other, but it is structured to incentivize you to turn on your partner, assuming that your partner is also turning on you.
What will the consequences be? #
Folks have been thinking about the progress towards AGI for some time, asking questions about whether or not a thinking AI would decide humans aren’t so useful and need to be eliminated.
But what about non-AGI agentic AI taking jobs? What do we need to think about in terms of the consequences here?
- If people can be replaced by agents, how does that impact the overall economy?
- Do those people just no longer have income?
- Do they just find other new careers? If so, where do those come from?
- If those people laid off no longer have income, then it’s assumed they’re also no longer purchasing at the same rate. How does that affect the overall p&l of the companies at large?
- How does this make the concerns of “too much money in the hands of too few” problem worse or better?
Additionally, it’s fair to think that the companies that have the resources to build AI Agents are exclusive to larger established companies. This puts mom and pop shops, or start-ups, at a disadvantage. It is worth questioning whether new companies will be able to clear the resource hurdle in this space to become successful at all. And if they can’t, does that mean we enter an even larger form of corporate controlled government going forward?
As always, it’s an interesting time to be alive. Stay curious everyone!