History's Future  /  Essay № 007  /  Part II of II

The
Floodlight

What AI does to the seesaw — and why the singularity is a coastline that never stops growing.

A machine has arrived that drinks from the jar of knowledge a billion times faster than any of us. Does it finally collapse the uncertainty between what we know and what we don't — or does it do something far stranger to the shadow behind the torch?

BY ASHOK MEHAN PUBLISHED JULY 2026 PART II OF II 5 MIN READ

Somewhere along that hard road, I picked up an unlikely, genuine fluency in physics, cosmology, and technology — the kind you earn by lessons rather than lectures. Occasionally I would surprise friends at dinner parties, along with myself, when I could explain the Planck constant without stumbling into an abyss of dark energy — the kind of abyss you get when thoughts of a college crush come wafting through your brain from nowhere, uninvited, mid-sentence, and you stand there gaping, trying to remember what it was you were saying about quantized energy.

I mention this because the fluency was earned one sip at a time. Every fact I learned cost me an evening; every evening revealed two more evenings I owed. That was the deal. That was always the deal, for everyone, since the first human wondered what the stars were.

The deal just changed.

Enter the Machine

Artificial Intelligence is about to change the knowledge problem for almost every subject — and it is doing so precisely by shifting what "knowledge" even means. The old paradigm rewarded the accumulation of facts. The new one rewards something else entirely: the ability to acquire the right knowledge on demand matters more than hoarding knowledge for its own sake. The jar now refills faster than any human throat can drink.

So the natural question, the one Part I ended on, is this: does AI collapse the knowledge–ignorance seesaw, finally letting us have both position and momentum at once? Or does it amplify it?

My answer, after a year of reading the research at midnight when I should be sleeping, is that AI is the first technology in history that expands both sides of the seesaw at the same time — and that this is the whole story of the singularity in miniature.

The Jagged Frontier

Start with the answers-arrive-instantly side, because it is real. In a Harvard Business School and Boston Consulting Group field experiment involving 758 consultants, workers using GPT-4 on tasks inside the tool's competence completed 12.2% more tasks, finished them 25.1% faster, and produced work rated more than 40% higher in quality than a control group.[1] The authors gave the shape of AI's ability a wonderful name: the "jagged technological frontier." Inside the frontier, the machine is a superpower.

−19pp
drop in correctness for consultants using AI on a task chosen to fall just outside the tool's competence — because the machine answers every question with the same serene confidence whether or not it has any business doing so.[1]

That is the shore of ignorance from Part I, rebuilt in silicon.

What the Machine Does to Us

The research on what AI does to our side of the seesaw is more unsettling still.

A Microsoft Research and Carnegie Mellon survey of 319 knowledge workers, presented at the CHI 2025 conference, gathered 936 real-world examples and found that higher confidence in the AI was associated with less critical thinking, while higher confidence in oneself was associated with more — trust in the machine trading directly against scrutiny of its output.[2]

Worse, a study out of Aalto University with the marvelous title "AI Makes You Smarter, But None the Wiser" found that AI use flattens the famous Dunning–Kruger effect and, in a genuinely eerie twist, reverses it.[3] As the senior author Professor Robin Welsch put it, "when it comes to AI, the DKE vanishes. In fact, what's really surprising is that higher AI literacy brings more overconfidence." The people who felt most fluent with the machine were the least accurate at judging their own performance.

And an MIT Media Lab study wired 54 people to EEG machines while they wrote essays with ChatGPT; lead author Nataliya Kosmyna reported that the ChatGPT group showed the least brain connectivity — and afterward the AI users could barely quote the work that had appeared, minutes earlier, under their own names.[4]

Do you see what all of these have in common? AI does not shrink your ignorance. It hides it. It hands you a polished answer so quickly and so fluently that the shadow never gets a chance to grow — the torch has been replaced by someone else's floodlight, and you never learn where the dark was. The seesaw does not disappear. Your side of it just gets quietly disconnected.

AI does not shrink your ignorance. It hides it — the torch replaced by someone else's floodlight, so you never learn where the dark was.

The Coastline-Manufacturing Machine

And yet — and this is the disturbing-but-wonderful flip that keeps me hopeful — on the frontier of discovery, AI is doing the exact opposite. It is generating known unknowns faster than any generation of humans in history. Point a model at a protein or a proof or a hundred million data points, and it does not close the questions; it opens them, at a resolution the unaided mind could never reach. Every answer it accelerates arrives holding three new questions by the hand.

The physicist Marcelo Gleiser, in his book The Island of Knowledge, argues the appropriate response to progress is not arrogance over the territory conquered but humility before the coastline that keeps getting longer.[5] AI is a coastline-manufacturing machine.

Where I Land

The singularity thesis of this whole series is that dramatic, irreversible change is not coming — it is already here. And the uncertainty principle is how I know it. AI is the first technology that expands both arms of the seesaw at once: answers arrive instantly, and the frontier of what we don't know grows faster than it ever has. The jar refills at a speed we have never seen, and it opens new compartments faster than we can count them.

Which returns me, inevitably, to the unanswerable question from Part I — is it better to have knowledge, ignorance, knowledge of your ignorance, or ignorance of the knowledge required? — except now the machine is answering the first part for us at a billion words a second, and leaving the last three entirely, terrifyingly, gloriously to us.

That last part is not a footnote. It is the whole assignment. The machine can flood the island; only you can decide to keep walking to the water's edge. So use AI to expand the island, but go and stand on the shore yourself — verify what it hands you, ask the question behind the question, and treat every instant answer as an invitation to find the next unknown rather than an excuse to stop looking. Ignorance you can see is a gift. Ignorance a floodlight has hidden from you is the only kind that can actually hurt you.

For someone who reads quantum mechanics at midnight and thinks about consciousness in airport lounges, that is not a problem. It is the entire point. The universe is inexhaustible. There is always a next question. The shore never stops growing — and I, for one, intend to keep walking the beach.

I'm working on it. You're reading part of the research. Come back in three days — there will be more.

Sources & Further Reading
  1. Dell'Acqua, F., McFowland III, E., Mollick, E., et al. (2023). "Navigating the Jagged Technological Frontier." Harvard Business School Working Paper No. 24-013; published in Organization Science (2025). 758 consultants; within-frontier tasks +12.2% complete, +25.1% faster, +40% higher quality; ~19pp drop in correctness outside the frontier.
  2. Lee, H.-P., Sarkar, A., Tankelevitch, L., et al. (2025). "The Impact of Generative AI on Critical Thinking." Proceedings of the 2025 CHI Conference (Microsoft Research / Carnegie Mellon). 319 workers, 936 examples. DOI: 10.1145/3706598.3713778.
  3. Fernandes, D., Villa, S., Nicholls, S., et al. (2026). "AI Makes You Smarter, But None the Wiser." Computers in Human Behavior, 175, 108779 (Aalto University). DOI: 10.1016/j.chb.2025.108779.
  4. Kosmyna, N., Hauptmann, E., Yuan, Y. T., et al. (2025). "Your Brain on ChatGPT." arXiv:2506.08872 (MIT Media Lab). 54 participants across Sessions 1–3.
  5. Gleiser, M. (2014). The Island of Knowledge: The Limits of Science and the Search for Meaning. Basic Books.
  6. Wheeler, J. A., quoted by John Horgan in "The New Challenges," Scientific American, 1992, Vol. 267, No. 6 — the "island of knowledge / shore of ignorance" image referenced in Part I and throughout.
← Essay 006: The Seesaw of Uncertainty (Part I) All Essays ↑