Let’s be honest about where we are. Hiring freezes everywhere. Headcount reductions masquerading as “AI optimization.” An administration working overtime to make everyone—Republicans, Democrats, and Libertarians—equally miserable. Economic instability that makes long-term planning feel like a joke.
Meanwhile, we’re all being sold the dream: AI will make you more productive! You can do more with less! Just automate everything!
But if you’ve actually tried to implement AI beyond the otter-using-wifi on-a-plane demos, you know the reality is messier. Sure, some people claim amazing productivity gains, but notice how they’re usually the ones trying to sell you AI solutions.
The truth is, most organizations still need human intelligence to make AI work. The question I see execs asking is how to spend your extremely limited headcount budget to get maximum return.
And that’s where my completely counterintuitive strategy comes in: stop competing for expensive CS graduates and start hiring philosophy majors.
The Institutional Knowledge Trap
Before we get to the solution, let’s acknowledge what’s happening when companies fire people to “make room for AI.” You’re not just eliminating a salary line. You’re destroying a repository of competitive advantage: someone who knows how things actually get done, who has the relationships to cut through bureaucracy, who remembers what failed spectacularly three years ago.
Each employee represents years of accumulated institutional knowledge that AI can’t replicate. Sure, fire people who genuinely suck—that’s just good management. But don’t confuse “expensive” with “replaceable.”
The Bandwidth Problem
Here’s what most executives miss: your team is already running at 100%+ capacity. In tech, 50-80 hour weeks are standard. There’s no bandwidth for the exploration and experimentation that effective AI adoption requires. You know why people won’t admit they are using AI? They are afraid you’ll give them more work.
You cannot expect people to figure out AI integration while they’re drowning in their current workload. If you want AI to actually deliver productivity gains, you need to create space for thoughtful implementation. And given current headcount constraints, that means being strategic about who you hire.
The Counterintuitive Hire: Liberal Arts Graduates
Here’s my completely insane idea that more companies should consider: stop competing for CS graduates and start hiring philosophy, history, and English majors.
At Stanford, all the recruiters are fighting over the data science, CS and AI students. Meanwhile, there are brilliant critical thinkers graduating from humanities programs who don’t know what to do next. They’re considering consulting or law school because they need to pay rent.
These graduates are:
- Significantly cheaper than CS hires
- Trained in complex analysis and abstract thinking
- Skilled at breaking down complicated systems
- Available (because everyone else is chasing the same technical talent)
- Desperate for opportunities that don’t involve more student debt
The Strategic Implementation
Here’s how to make this work within your constrained budget:
1. Hire the analyst, pair with the expert. Take that smart philosophy graduate and partner them with your senior marketing person or lead engineer. Their job isn’t to build AI systems—it’s to understand your business deeply enough to identify where AI makes sense and where it doesn’t. And get them a Maven class or subscribe to Linkedin Leaning or Coursera so they can quickly get up to speed on how AI works. We live in a time of ridiculous riches of quality education on AI. A recent grad is an expert at learning. Unleash them!
2. Map your workflows first. Most businesses become giant organic tangles where nobody really knows what anyone does to get things done. Have your new hire systematically document every workflow and use case in your organization. This alone will probably save you money by identifying redundancies and inefficiencies.
3. Assess risk tolerance for each use case. AI is a magical, marvelous making-shit-up machine, and that’s a feature, not a bug. But you need to know when making shit up is acceptable and when you need facts. For each potential application, ask: What’s the cost if it hallucinates?
A PR release with made-up facts is embarrassing. A customer service response with wrong information could be costly. A financial calculation could be catastrophic. Where can that creativity on tap be used, and when should it be avoided?
4. Design human oversight accordingly. High-risk applications need human fact-checkers. Low-risk creative tasks can run with minimal oversight. Medium-risk applications might need RAG systems or other validation approaches. The key is matching the oversight level to the actual risk, not just implementing AI everywhere because it’s trendy.
5. Invest in your new strategic asset. After six months, you now have someone who deeply understands your company, knows where AI fits (and where it doesn’t), and has become invaluable to your strategic growth. You started with a cheap liberal arts graduate; you now have a strategic AI implementation expert who costs less than half what you’d pay to hire that expertise externally. Give them a raise. Nothing creates radical loyalty like an unasked-for bump in salary.
The Reality Check
We’re dealing with immature technology, regardless of what foundation companies claim. AI produces incredible results one minute and blatant lies the next. The model you choose, the post-processing, the context handling—everything matters. It makes a huge difference which foundation company’s model you use, how you handle the post-processing, how you manage context.
But that doesn’t negate the fact that this is a marvelous, magical making-shit-up machine. And we have to know when making shit up is the right choice and when we just want facts.
This isn’t replacing humans anytime soon. It’s augmenting them, but only if you give them the bandwidth to figure out how—and only if you’re willing to accept a certain amount of risk.
The Bottom Line
If you want AI to actually drive productivity in your organization rather than just burning through your budget, you need to invest thoughtfully in people who can integrate it intelligently. Given current economic constraints, that means being creative about talent acquisition.
The companies that figure out how to systematically identify AI use cases, assess risk appropriately, and implement human-AI workflows will have a massive advantage. The ones that just throw AI at problems without understanding their own workflows will wonder why their productivity gains never materialized.
And the ones that fire people to “make room for AI” will discover they’ve eliminated the institutional knowledge needed to make AI actually work.
In a world where everyone’s competing for the same expensive technical talent, the smart move might be hiring the brilliant humanities graduate who can think systematically about complex problems. They’re cheaper, available, and might just be better at figuring out where AI fits in your business than someone who only knows how to build models.