Lynn Raebsamen took the stage (and the audience’s expectations) with a charming mix of realism and irony about the current state of artificial intelligence. Contrary to the messianic promises plastered across tech ads and keynote speeches, she argues AI hasn’t quite become the omnipotent overlord—or automated colleague—everyone’s either fearing or banking on.
Raebsamen shares a personal aha moment that inspired her to write Artificial Stupelligence, and reframes the entire conversation (watch the video to catch it.)
Here are the main takeaways:
🧠 The Only Thing Intelligent About AI is…
Raebsamen pulls back the curtain on the AI hype machine, suggesting that:
- The most intelligent thing about AI today is… the marketing department.
- Much of the public fear and industry panic? It’s often manufactured by the very companies hoping to sell AI as a miracle fix.
- Underneath the glossy demos and buzzwords lies a truth that’s more sitcom than science fiction: AI still very much depends on humans.
👷 Human Beings: Still Hiring, Still Required
Despite headlines screaming “robots are taking your jobs!”, Raebsamen notes the recent tech job cuts likely aren’t a byproduct of AI efficiency. Rather, it’s a case of Silicon Valley indigestion—too much hiring, too fast, and now trimming the fat.
- AI isn’t replacing us; if anything, some companies are just hitting reset on their bloated payrolls.
- And ironically, you often need more humans to build and train AI—especially the kind meant to replace… other humans.
🏷️ The “L” in LLM is for Labeling (and Labor)
One of the most sobering, yet entertaining parts of the discussion centered around the glamorous world of data labeling—or as Raebsamen puts it, “low-paid, invisible labor from developing countries.”
- While futuristic narratives love to imagine AI auto-labeling and learning at sci-fi speed, in reality, AI still needs us to hold its data-laden hand.
- Without human-tagged data, AI models start serving up surreal soup—random facts mixed with confident nonsense. This is the source of hallucinations, not digital daydreams.
The catch?
As AI models grow more powerful, the real work shifts to training them on accurate, richly contextualized data—created, guessed what, by people.
🧞♂️ Your Three AI Wishes: Be Careful What You ~Prompt~ Wish For
Despite sensational talk of “AI evolving past humanity,” Raebsamen grounds the conversation:
- AI can’t learn without us—at least not yet.
- Building smarter models isn’t the real challenge. Feeding them real, relevant data is.
- And here’s the irony for the ages: You need human labor to teach the AI to eliminate… human labor.
That’s not automation, that’s bureaucracy with extra steps.
🎤 Final Thoughts
Raebsamen isn’t AI-phobic—just AI-realistic. She believes in the power of the tool but isn’t sold on the pipe dream. Her message is humorous but pointed: AI still has a long way to go before it can actually replace us. In the meantime, let’s not lose our heads—or our jobs—over a predictive text engine with a God complex.
For more insights about what AI can or cannot do, check Raebsamen’s book “Artificial Stupelligence: The Hilarious Truth About AI”:
