Mathematicians issue warning as AI rapidly gains ground
science.org293 points by pseudolus 4 days ago
293 points by pseudolus 4 days ago
> Mathematics produces not only a body of results, but also understanding, clarity, and judgment among the communities of mathematicians who have shaped them, often in the context of their own autonomously guided research. This expert knowledge is essential, both to effectively use mathematics, and to continue to articulate new and significant research questions.
In a word, the job of the mathematics department is not only to produce mathematics, but mathematicians.
Similarly, the output of programming is not only a program, but also a programmer. It is you.
Outsourcing the work deprives you of who you become by writing it.
> Outsourcing the work deprives you of who you become by writing it.
Just because AI can do something that resembles work should not mean outsourcing work to it. Mathematicians should not outsource their work to AI just like programmers should not outsource programming to AI.
Humans working with AIs in a tight loop means intellectual work becomes more high-level and creative, but a human should always own the work, validate it and stake their reputation to it. Simply ban any humans who produce low quality work using AI.
I don't think that is possible. Humans have always taken the path of least resistance, especially when it comes to work/school.
The idea that we just "trust everyone to carefully check and learn from AI output" as our barrier to human skillsets eroding is never going to work.
There is an Anthropic engineering post on HM front page that addresses this exact issue:
"... supervise the agent’s behavior via a human-in-the-loop. Claude Code previously protected against agents taking unintended actions by asking users for permission at each turn. Theoretically that works, but we’ve found the approach to be fallible. Our telemetry showed users approved roughly 93% of permission prompts. The more approvals a user sees, the less attention they pay to each, becoming over time much less diligent in their supervision. "
https://en.wikipedia.org/wiki/Normalization_of_deviance
Yeah I'm seeing this with the attitude towards AI. Especially as the economic benefits increase, we will justify increasingly reckless approaches. (Probably until some major catastrophe. That seems to be how these things go.)
While it is human nature to minimize energy expended on doing things, progress has always come from the minority who prioritize disciplined thinking and action.
While minimizing energy spent worked well in historic periods where survival was hard, in this era of abundance and a complex, interconnected and fragile civilization, the same instinct becomes harmful.
> Mathematicians should not outsource their work to AI just like programmers should not outsource programming to AI.
There is a huge difference between the two. Mathematicians work on discovering fundamental truths of the universe that go into the corpus of human knowledge forever. Programmers create utilities.
I know a PHD in math that claims math is invented by ourselves and not any universal true. So well, depends who you ask. Also some of these programming utilities may outlive some math proof. Time will tell
Of course they are universal truths. We may have made up the rules/abstractions/symbols to represent the underlying but a proof will hold in any part of the universe. Infact, math will hold in any universe. You could change every fundamental physical property of the universe and those proofs will still hold.
Those are wild claims that you can't possibly prove. They are typically assumed to be the case to the extent that we even think about them but in the end are largely unanswerable philosophical questions.
It’s not a claim, it’s a pretty self apparent fact. To wrap your head around this, as the simplest example 2+2=4 doesn’t change anywhere or under any different physical law. It’s as universal as you can get. There’s nothing philosophical about this.
> 2+2=4 doesn’t change anywhere or under any different physical law.
How about python3:
>>> input() + input()
2
2
'22'
>>> .2 + .2 + .2 == .6
FalseIt is a claim, and you can't test it. If physical laws varied between galaxies you wouldn't know unless we were able to measure it. So the current bound on physical phenomena is whatever the resolution of our observational data is, coupled with our models that match it.
How are you going to get observational data for a different universe? Does such a thing even exist? What is its nature? You're operating well outside the bounds of human knowledge.
What you are actually saying there is that you can't imagine 2+2 being anything other than 4. That's perfectly reasonable but it's not the same thing.
There is no circumstance where 2+2 does not equal 4. It is a literal fact.
At the most fundamental level, you can only have a discreet or a non-discreet universe. If it’s discreet, there are countable things and 2+2 = 4 is true. In a non-discreet universe there are no countable things, but the universe itself is countable. If the universe were non-discreet and infinite, you could still count the infinities so it’s still true.