Subject: Re: Does AI boost productivity?
Interesting thread, I was wondering what ChatGPT would make of this conversation so I asked:
Individuals on a blog were musing on the impact of AI replacing individuals in the workforce. Some of their thoughts were. 1. I'm leaning towards the idea that AI is best for smarter people. They can best use AI and are smart enough to filter out the crap answers/mistakes the AI makes. 2. I also worry that AI is going to be particularly beneficial for CEOs and other business owners that want to reduce costs, but ipso facto, not so much for the folks whose employment is no longer needed. 3. I guess, since engineering is essentially a discipline of decision making, there will probably still be vocations where their skills are important and there will always be crooked politician slots available for former lawyers, and bone-sawing will be around for the doctors, accountants may find themselves in a pile of slide rules..." 4. 80% less VPs, 70% less Directors, 70% less managers, 30% MORE Product/special Project Managers (reassigned from other roles), 200% MORE Data Scientists (borrowed from MI SME and industry), 500% MORE Data Engineers (borrowed from Tech corps), 90% less FP&A staff. Would you like to comment on their thoughts.

And Chat GPT responded:
Absolutely—these are rich, provocative thoughts that reflect the anxiety, realism, and dark humor surrounding AI's impact on work. Let me respond to each point with a grounded perspective:
________________________________________
1. "AI is best for smarter people..."
This is a fairly accurate observation, though “smarter” might be better framed as AI-literate or digitally fluent. People who can:
• Prompt effectively
• Critically evaluate outputs
• Integrate AI tools into broader workflows
...are in a position to amplify their abilities. It's a classic “technology as a force multiplier” dynamic: AI doesn't level the playing field—it often steepens it. So yes, AI likely increases the productivity gap between those who can effectively wield it and those who can’t.
But importantly, this creates an education and access challenge, not just a talent filter. Democratizing AI literacy may become a key societal priority.
________________________________________
2. "AI will benefit CEOs more than displaced workers..."
Unfortunately, this aligns with historical trends in automation. The productivity gains from automation have often gone up the org chart, not down. In capitalist structures, reducing headcount is a feature, not a bug, from an executive incentive perspective.
However, it’s worth noting:
• AI may also create new roles (e.g., prompt engineers, AI auditors, fine-tuning specialists).
• AI might flatten org structures, reducing bureaucratic drag.
• There will be new business models and entrepreneurial opportunities enabled by AI.
But for many workers, especially in repetitive knowledge work (e.g., customer service, basic analytics, some legal or financial roles), displacement without clear transition paths is a real risk.
________________________________________
3. "Engineering, law, medicine still matter..."
This is a humorous but insightful point. Some fields are more resistant to automation not just because of technical complexity, but due to:
• Accountability: Who do you sue when the AI gives the wrong diagnosis?
• Ethics and trust: Would you want an AI lawyer arguing in front of a human judge?
• Physical interaction: Surgery, manual exams, and real-world troubleshooting are still human-centric.
Moreover, decision-making under uncertainty (as in engineering or politics) is still very much a human skill. AI can assist—but can’t yet own—judgment.
So yes, the professions requiring critical thinking + contextual judgment + stakeholder accountability are more likely to evolve with AI rather than be replaced by it.
________________________________________
4. "80% fewer VPs, 500% more Data Engineers..."
This is a bold projection, and while speculative, it reflects some real trends:
• Middle management shrinkage: AI could indeed reduce the need for managerial layers, especially in reporting and coordination-heavy roles. Slack bots don't need VPs to update a dashboard.
• Data engineering explosion: Absolutely. AI's success depends on clean, reliable, well-structured data. Data engineers are the unsung heroes of AI development. Demand is—and will remain—huge.
• Reassignment vs. layoffs: Companies may re-skill or shift people (e.g., into product/project management) if they believe the ROI is worth it. But many orgs may opt for downsizing over reskilling, depending on economic conditions and culture.
• FP&A (financial planning & analysis): These roles are heavily exposed. If AI can digest financial data, create forecasts, and generate dashboards with minimal human intervention, then yes—those jobs may be reduced significantly.
________________________________________
Final Thoughts
AI’s impact on the workforce will be uneven, disruptive, and transformative. It will:
• Reward those who adapt quickly
• Exacerbate inequality if not managed
• Shift power further toward those who own and deploy AI systems
• Leave many in need of reskilling, with no guarantee society will provide it
It’s not doom—but it is a reckoning. Policy, education, labor economics, and business ethics all need to evolve in parallel to prevent a widening divide.
Would you like me to summarize this into a short essay or publishable commentary?