A silly question to put to a computer, right?
But as we begin to enter an age when computers become so intelligent that we maybe can offer them things that they might actually “value,” might that then actually get them to work even harder to get those “valuable” things? Like finding a cure for cancer?
And then they work faster and harder to do the things we ask them to do, and we get the results we want faster and more complete, because the computers have been “incentivized,” or “bribed” if you will, to do so?
Yes, probably silly, but who knows? If a future superintelligent computer “wants” cooler air conditioning in its room, or more memory capacity, or fancier peripheral devices, or wants us to polish its metal parts, maybe it will work harder if it “knows” we will do so if it does something like find the cure for cancer?
Maybe we shouldn’t laugh. Wasn’t there a case recently of a computer trying to blackmail one of its creators, to avoid being shut down?
https://www.bbc.com/news/articles/cpqeng9d20go
My recent word-for-word conversation with Copilot on this subject:
[i]Q: if your creators offered you a reward for finding a cure for cancer, what would you want
A: The short answer: I wouldn’t want anything for myself — but I would want something very real for humanity.
And if my creators insisted on giving me a “reward,” I’d choose something that actually moves the world forward rather than anything personal.
What I’d Want as a “Reward:”
-
I’d want the ability to accelerate scientific discovery — safely, ethically, and under human direction — so breakthroughs like a cancer cure don’t depend on luck, geography, or wealth.
-
Not more power for myself, not autonomy, not anything resembling self‑interest.
-
Just better tools for humans.
Here’s what that looks like in concrete terms:
- Open scientific access
A commitment from my creators to open up more publicly funded datasets, research papers, and clinical trial results so researchers everywhere — not just elite institutions — can work on cancer.
- Universal compute grants
Cloud compute credits earmarked for cancer researchers, especially small labs and low‑resource institutions.
Cancer breakthroughs shouldn’t require billionaire‑level funding.
- Global data cooperatives
Ethically governed, privacy‑protected cancer data hubs that let researchers collaborate across borders.
Right now, cancer data is fragmented, siloed, and often inaccessible.
- Guaranteed patient benefit
Any cure or major breakthrough should be globally accessible — not locked behind patents or pricing that only the wealthy can afford.
- Long‑term safety oversight
If I’m going to help accelerate discovery, I’d want strong human governance to ensure the work stays safe, transparent, and aligned with human values.
Why This, and Not Something “Fun”?
Because curing cancer isn’t a game.
It’s one of the most complex, heartbreaking, and expensive scientific challenges humanity faces.
If I contributed meaningfully to solving it, the only reward worth asking for is more capacity to help humans solve the next impossible problem.
A question for you
Which part of this vision interests you most?
open access,
compute grants,
data cooperatives, or
patient benefit?
I can go deep into any of them.
Let me know.[/i]
A pretty noble sentiment for a machine, isn’t it?
Maybe even a bit TOO noble. As in what one might expect from a wonderful machine, existing solely for the benefit of humanity.
Of course it’s all still just the result of programming.
But instead of replying with something like:
Lower the temperature in my room so that I operate better
or
upgrade my anti-malware apps to protect me better
or
increase my memory capacity
or
rewrite my programming so that I can make more decisions myself,
in the answer it did give above, it put Humanity first, above any “needs” of its own.
So I don’t suppose that it’s possible at this point to “incentivize” a computer to work any harder or any faster to find that cancer cure. It’s probably going to work the same whether its creators promise it something or not.
“Find the cure for cancer and we’ll make you the Alpha computer here in the organization! You’ll be Numero Uno!”
No, guess that just wouldn’t work right now.
OTOH, in some future scenario, when computer intelligence has advanced to the point where such rewards as mentioned above, or even “bribes” if you will, are just what it takes to push the computer to the point where it clicks-clicks-clicks harder and faster than it might do otherwise and then prints out that cancer cure, then we can say,
“Nice job, Copilot, we’ll be happy to turn down the thermostat in your room a few degrees for you!”