fn-mote

Today I foolishly thought maybe I was the one who was holding it wrong […]

Apparently goes on to describe being confused about parallel reading from the standard input?

The citation needed is for the other part, which claims a mathematician got an amateur proof which was correct.

Nobody is reading unsolicited proofs. They are like spam.

Source: personal friend of a “crank”.

The concern is that there could have been expert knowledge input, whose importance/worth we are unable to evaluate.

I don’t believe a mathematician produced the counter example secretly, but how much did they contribute to the result?

AI isn’t magic, so to evaluate the value delta, you need to know the value of the input.

On one hand, yes.

The parent wants to be skeptical… nothing wrong with being suspicious of marketing claims, right?

I would be a lot less impressed if I found out the session was guided by an expert in the field who already had a good idea of where to look. For example, I don’t believe that the results Terrance Tao gets from an LLM are comparable to what I am going to get.

I’m not even saying I’m a skeptic. Just that there’s nothing wrong with keeping your eyes open and asking for details.

to get good enough that choice of shoe matters

Bowling shoe vs running shoe? As soon as you’re trying to get better, the shoes matter a lot. They let you use the established techniques.

Bowling shoe vs socks? Never tried it, but I’m going to guess once you can break 100 that socks don’t cut it.

more absurd to think that humans are near the top of that curve rather than right at the bottom

The “absurd” dimension does not enter. This is a situation where you have no evidence at all.

In the absence of any information, the average (mean or median) is your best guess. Now where that average is, you have no idea.

There might be a thing beyond intelligence that we can't even conceive of.

This statement already supposes there is a thing called “intelligence”. People have been pretending to measure this for more than a century. Modern thinking at least says what we call intelligence is not a single concept.

How would this be different from taking a yoga class from a video?

I would be willing to be proven wrong, but I doubt the ability of LLMs to give useful corrections in yoga much more than their ability to write useful code.

The customer has more power than a patient.

I definitely want to be a customer.

I think this is an unfair take.

The necessity of teaching nurses that doctors orders are not sacrosanct comes from the bitter experience of doctors giving orders that are wrong.

Asking for clarification is great, but doctors can be very reluctant to hear. The bottom line is that the nurse must not do certain things and the certification exam is there to make sure they know it.

Think of it in relation to the “anybody can stop the assembly line” part of quality control.

with no option to say "Actually they did their best but this isn't their fault."

In this case, the lack of such an option is obviously a flaw in the assessment system.

How to fix that? Major political issue, I suppose.

When LLMs stop helping me with my work, I’ll stop going to doctors who use them.

Too bad for me, it’s a real mixed bag. I need a doctor who is an AI-using LLM skeptic, I guess.

Ed: I disagree. My recollection is that machine learning was routinely sold as “AI” even when it obviously wasn’t. (IBM’s Watson was good at Jeopardy but not real medical applications.)

This isn’t exactly the same, but nothing in the book Paradigms of Artificial Intelligence would be considered AI today.

AI-powered alerts from their apple watches that detected cardiac events

Surely these are “good old-fashioned AI” (statistical learning) and not LLM, though.

I just want to be clear that the “medical LLM” tools are the new ones, and the Apple Watch alerts aren’t.

AI is a force multiplayer and scarily dangerous in the wrong hands

You have to spell this out a lot more if you want to have credibility.

I’m not seeing anything in discussed here that seems scary.

AI hasn't done anything we couldn't already do. It's just doing it faster and with more mistakes.

You forgot CHEAPER (at least now, burning VC money), which is a major motivating factor.

I really don’t think that’s what’s going on here.

People interact with AI, talking to it like a human. Of course they start to believe it’s rational like a human.

LLM does all of the entry level tasks better than the students. Partially because the answers are in the training set, and partially because it has gotten that good now. Hard not to start to believe it is “competent”.

I personally have had a real hard time getting traction talking about making sure the way we assess AI is not based on material it has trained on. YMMV as always, but I think the large training corpus contributes to the (unreasonably) high level of faith in the machine.

I would be so much happier with this comment if it gave any detail about what was worse.

As it is, it’s just a rant not a contribution to dialog.

It’s not more important information than a password manager, it’s just more.

You can just not have phones, but it's socially isolating.

I hear this plenty, but kids are literally forced to go be with their peers (hopefully including some friends) for 6+ hours a day (in my country). Don't they have enough time to arrange their social lives in person?

It’s not as bad as phone apologists say.

Pretty sure iOS is providing this? At least for messages and voicemail.

Curious if anyone can explain the Shodan packets described here.

This is barely an issue… changing the redirect is instantaneous.

Can't the customer rate-limit connections?

Bad guys use a botnet (compromised residential computers and routers) to defeat rate limiting and ip-based blocking.

(2022)

I can only imagine what it’s like now.

The difference is, if you tell the LLM that the component is too big and ask it to use some strategies to break it up, it will.

Humans? Also maybe, but I always suspect the humans actually don’t know a better way. The LLM does, just that pathway wasn’t activated.

My main argument about this kind of work is that it is (essentially if not exactly) in the training set.

I want a different argument before I believe that LLMs are doing “out of training dataspace creativity” (extrapolation not interpolation).

Depends on your organization I suppose.

Some places you see people pushing their Shiny New Rewrite, basically as resume fodder or promo packet material.

I describe that kind of decision as political because it doesn’t seem to come with a legitimate cost-benefit analysis.

Advice: look for a job, then use what you learn you need in the job to motivate your next studying.

If you write a PhD and feel like it was a waste, too bad. I think if you had approached it differently, you would have gotten something else out of it. By that point you’re supposed to be more responsible for your own learning.

No comment on the advisability of the degrees.

Use Anki

Anki is great for studying, but the card creation experience sucks. To be specific: I found creating any custom card type immediately dropped me into the bowels of CSS. It felt like writing HTML by hand.

Is there any facility for re-using shared pieces?

I felt like it needed a static site generator type tool to move up a layer of abstraction and reduce the copying of chunks into my card. Is there one? Please mention if so.