Ask HN: What might Aaron Schwartz have said about AI today? | Hacker News Reader

Ask HN: What might Aaron Schwartz have said about AI today?

35 pointsabtinf20 comments

I only had the pleasure of meeting Aaron once, at the YC open house after the first startup school in Cambridge. I was pitching sort of a competitor to Infogami and he helpfully whipped out his Sidekick and showed me a bunch of stuff. For the first and only time in my life, I was immediately struck by the thought of “now this is a kid who understands things.” His later work only reaffirmed my view and, though I could only watch from afar, he was critical to building a different kind of world. His blog was always insightful and a source of value.

Often, since the announcement of ChatGPT, I’ve wondered “what might Aaron have thought about this?”

Perhaps those of you who had the good fortune to know him better might share anything he might have said about AI or knowledge silos or the nature of information or free will or anything related?

View on HN

“Before I went to college I read two books. I read a book “Moral Mazes” by Robert Jackall which is a study of how corporations work, and it’s actually a fascinating book, this sociologist, he just picks a corporation at random and just goes and studies the middle managers, not the people who do any of the grunt work and not the big decision makers, just the people whose job is to make sure that things day to day get done, and he shows how even though they’re all perfectly reasonable people, perfectly nice people you’d be happy to meet any of them, all the things that they were accomplishing were just incredibly evil. So you have these people in this average corporation, they were making decisions to blow out their worker’s eardrums in the factory, to poison the lakes and the lagoons nearby, to make these products that are filled with toxic chemicals that poisoned their customers, not because any of them were bad people and wanted to kill their workers and their neighbourhood and their customers, but just because that was the logic of the situation they were in.

Another book I read was a book “Understanding Power” by Noam Chomsky which kind of took the same sort of analysis but applied it to wider society which you know we’re in a situation where it may be filled with perfectly good people but they’re in these structures that cause them to continually do evil, to invade countries, to bomb people, to take money from poor people and give it to rich people, to do all these things that are wrong. These books really opened my eyes about just how bad the society we were living in really is.” ― Aaron Swartz

I guess he might start with "AI would at least spell my name right." https://en.wikipedia.org/wiki/Aaron_Swartz

You may be interested in Lex Fridman's interview with Eugenia Kuyda: https://lexfridman.com/eugenia-kuyda/. Among other things, they discuss what Eugenia did with machine learning in response a close friend's death, in an attempt to preserve him and conversation with him.

Most public intellectuals are not petty enough to chastise people for misspelling or mispronouncing their names. (I've noticed this recently watching various interviews with on current events with people who have Hebrew and Arabic names.) Idk why Swartz would be any different.

True, that's my gripe.

>Most public intellectuals are not petty enough to chastise people for misspelling or mispronouncing their

Public intellectuals and pettiness is hardly a rare combination.

The Schwartz is strong with OP.

It’s crazy to think that if he would have just made a training set of JSTOR documents for a large language model he might have skirted around the copyright issues entirely.

I never met him in person, but I wanted to use an rss script he wrote, so I just emailed him as a random person from the internet. He quickly responded and added authentication to the rss software he was distributing on his blog. He was still in high school at that point before he became famous.

What an awesome guy! He will be missed.

Unfortunately you either die a hero, or live long enough to see yourself become the villain.

Just look where his old buddies are now, start with the Reddit folks.

He didn't not become a villain just because he died early.

He'd have stayed a hero, even later - like many who continue to be heroes in older age. He was already very different in disposition and willingness to do stuff for good compared to his Reddit peers.

In the real world the quote mostly applies to people who were already merely superificially good younger. Some corrupt politician which once was participating in the student movement (when it was fashionable and advantageous for them), some billionaire founder that paid lip service to "don't be evil" when their company was just growing and needed good PR, etc.

People who were geniunely good, and not just doing token "good gestures", tend to stay good.

It's painful to remember that we have to go through this without him. Hard to tell what his approach would be. If I had to guess, I'd say he would do everything he can and mobilise the community to create free (as in freedom), open-source, decentralised AI, to benefit humanity.

Perhaps we should scrape all of his known public works and sayings to then feed into an AI bot that would tell us what he might have thought of any topic, much less AI.

Nobody will ever know

www.988lifeline.org

Trying to find a link to the story of Aaron et. al (with declared intent) generating fraudulent ScholarlyArticles, submitting them to journals, and measuring the journal acceptance rate.

I see US vs Aaron, but no link to the SchoarlyArticle about - was it markov chains in like 2007 - submission of ScholarlyArticles and journal acceptance rates.

I mean, a reddit submission with markdown from nbconvert is basically a ScholarlyArticle if there's review and an IRB or similar.

Not to nitpick, but it's spelled 'Swartz'.

Well, we're talking about someone who was in legal trouble for taking information from academic libraries that he thought should be free for all. He was willing to put himself at a great personal risk for the sake of making information free. So he would probably be critical of OpenAI for not being open enough, and would probably champion the cause for more and better open source models.

I am sure you could ask The-Bloke on huggingface or reddit. Making all LLM available for anyone with small GPU, CPU or alike.

Aaron would oppose all the LLM being concentrated in the hands of a few mega corporations, and hope that there would be more open source LLM that possibly form clusters and groups to make them available for anyone.

Probably similar to what his compatriots in S05 YC Batch, Emmett Shear and Sam Altman, think:

"Why You Should Fear Machine Intelligence"

https://blog.samaltman.com/machine-intelligence-part-1

this is a LINK to him talking about AI when he was in his early teen years. https://archive.org/details/AaronSwartzInterviewedByLewKochW...

Feed all this blogs through a LLM, then ask it.

Reply on news.ycombinator.com