Are Layoffs Evidence of AI Impact? Link to heading
People point to AI-related layoffs as evidence that the economy and society are fundamentally changing. This is mistaken, for three reasons.
1. PR is not ground truth Link to heading
Companies often claim that layoffs are due to efficiency gains from AI use, entailing the replacement of workers by AI. These claims are made by PR departments and salespersons who are “managing their reputation”, not trained scientists looking for truth. Yes, the CEO counts as a salesperson, not as a subject matter expert.
Layoffs motivated by shareholder value? AI disruption sounds much better.
Hired too many people? AI disruption sounds much better.
Poorly performing business that needs to cut headcount? AI disruption sounds much better.
There are good reasons for doing so from a business perspective. Lines of credit, investments and customer interest depend on reputation, trust and confidence in future success. It is much better to appear to be forward thinking, even if it’s not true.
2. Executives make mistakes Link to heading
There are a lot of qualified and very intelligent people who earnestly believe that AI will change the course of the future, beginning right now. They will gladly cut headcount and invest in AI in an attempt to secure an edge, but this is a speculative bet on the future, not rigorous planning based on certain evidence.
Given the novelty of the technology and the terrible state of evaluation research, it is really hard to tell whether AI improves or degrades efficiency of an organization when deployed at scale. Programming is a favorite for automation right now, so let’s use this as an example.
Will more code generated by AI make better programs that better solve our real-world problems? Historical evidence tends to question the value of increasingly large codebases. Current evidence on AI agent use appears mixed. Microsoft is certainly all-in on AI and now having critical code quality issues. Amazon Web Services had a “trend of incidents” with “high blast radius” following “Gen-AI assisted changes”.
I would like to cite some data here, but honestly so much of it is small in sample size or proprietary or anecdotal or otherwise questionable that I would have to write a full academic paper on this subject. Frankly, I don’t see this is a worthwhile use of my time at the moment, so I’ll anecdotally state that wait-and-see is the best approach.
There is a famous aphorism for this: predictions are difficult, especially when they concern the future.
Yes, intelligent people can believe all of the hype. Yes, they can be entirely wrong and lose their bet. Yes, people in their organization will suffer as a result.
3. Influential people make a lot of money selling AI hype Link to heading
I get a lot of AI hype in my feed from a lot of different people. It’s a personal pet peeve of mine to check whether people have a conflict of interest.
- There are, of course, the self-declared AI influencers who always include a line of “follow me to not be left behind in the future of AI or something” at the end of their posts.
- Most AI hype posts I see tend to be from startup founders who are selling an AI-related efficiency product
- Some lawyers and domain experts have known (probably lucrative) contracts with AI product companies, so I assume these posts are (indirectly) financially compensated
Do I have a conflict of interest myself, now that I founded a consulting company in the legal tech space? I like to think that I am selling my time, not my honesty, but let me know how this turns out.
Conclusion Link to heading
Is society fundamentally changing? Possibly. Not necessarily because of AI, but because people uncritically believe in AI hype. Those are two entirely different things.