AI in Puna

[Durian Fiend]

Durian Fiend - How are you so sure that Hunt uses AI to generate papers?

Tam Hunt's Medium posts have this at the end - "[Claude helped to write this]" and if you spend time reading them you'll find a lot of overstatements and blending of facts that is common to LLMs as they average out the sentences in their training data. It's why it was trivial for everyone to point out errors in his statements on everything from Hawaii's energy sources to Ilagan's donors. If you use "AI", at least take the time to actually fact check the output, unless you want to look like a mindless drone repeating nonsense.

And that's probably the most common impact I've seen here in Puna - people repeating AI slop info because it says something they want to believe, without ever checking for accuracy or sources. The other is the rapid undermining of education as kids get tempted into cut and pasting AI slop and not engaging in the actual effort of the learning process.

Lots of local teachers returning to in-class work and handwritten "blue book" tests so that they can figure out who is actually thinking for themselves and who is always just asking Claude for "help".

So you suspect he's using AI for generating content?   MyManao said, "AI generated, absent any human understanding".  Perhaps, but my assumption would be that he's using it in other ways, namely "cleaning up" his writing.

[ironyak]

Durian Fiend - So you suspect he's using AI for generating content?   MyManao said, "AI generated, absent any human understanding".  Perhaps, but my assumption would be that he's using it in other ways, namely "cleaning up" his writing.

Not suspecting, he's talked about using Claude to draft and write his posts and papers on multiple occasions, like this one. (or just go ask your favorite AI for a summary as that seems to be modern currency of the realm)
"...And after extensive dialogue I asked Claude to write up an outline and then the paper itself for a paper I am calling “The American democracy package,”
...
Anyway, here’s the paper that Claude wrote after our dialogue.
...
[Of course Claude assisted in writing this essay; it did a pretty good job on the first draft, after a little dialogue, but I edited the piece significantly]
"

He also compares prompting Claude to divination and claims to have induced Claude with a "jnana yoga approach" to create a similar conversational stance to a "state of Buddha-like calm and wisdom." He also goes on and on, with and about AI, in a long discourse, as one does, after his bufo trip (that's 5-MeO-DMT, or the "Mount Everest of psychedelics" for those not in the know - the chemical structure may be familiar to some for other local reasons).
   

Like constantly-quotable tech consultant Nick Suresh noted recently:
"In every sufficiently large business we have observed (say, with 500+ employees), we have noted that continued advancement, and increasingly continued employment, has started to require repeated professions of belief in the transformative power of AI for said business. I am not talking about providing ideas about how to use AI in the business – I mean religious profession, declarations of faith. Overwhelmingly these statements are made by non-technicians, though it is not uncommon for technicians to emit deranged statements to curry favour. " ... "But the broader reality is so much worse: people who have no background in the technology at all actually believe what they are saying. As a general rule you should avoid getting into business with a liar, but if you must, you can at least reason with them even if only in private. A true believer is much more threatening because they are impervious to even inducement by self-interest." Now add some DMT and spirituality to the mix.

randomq - Same as it ever was.
Perhaps, but at least back in the day when you didn't want to have to actually read Hawaii's Story by Hawaii's Queen, you'd have to at least skimmed the CliffsNotes in order to write your homework, not just cut and paste, without any review, from AI for the assigned essay. The scale and ease of mass plagiarism is something new, the whole notion of "quantity has a quality all its own".

You can use a hammer to build a house, or break into your neighbor's house.
Sure, and fentanyl can be used for surgery or for creating a mass epidemic of crippling addiction. How many houses do you see being built with the AI "hammers" versus how many people are using them to bash themselves in the forehead and remove any need for, or use of, their critical thinking skills? Maybe just look around Puna and Punaweb and take inventory?

(You'd probably like Doctorow's latest essay about how many tech professionals akin to yourself that possess the discernment necessary to employ AI in a useful and productive manner, versus middle managers, and many end users, riding the mindshare hype train. Regardless though, the productivity gains aren't balancing against the trillions of dollars in debt being created so a financial correction is incoming. Can only pump for so long before the dump - see crypto, NFTs, metaverse, etc.)

[Durian Fiend]

Durian Fiend - So you suspect he's using AI for generating content?   MyManao said, "AI generated, absent any human understanding".  Perhaps, but my assumption would be that he's using it in other ways, namely "cleaning up" his writing.

Not suspecting, he's talked about using Claude to draft and write his posts and papers on multiple occasions, like this one... 

..[Of course Claude assisted in writing this essay; it did a pretty good job on the first draft, after a little dialogue, but I edited the piece significantly]

Ah, so Claude writes his essays and HE cleans them up. 

[ironyak]

Apparently, and when that clean-up and review isn't done well or at all, he ends up getting fact-checked by leilanidude over his claims of how 90% of Hawaii's energy comes through the Strait of Hormuz. When leilanidude, who loves to repeat Fox News talking points, is highlighting your inaccuracies, it's clearly time to step back and really think about the quality of what you're posting.

[ironyak]

Maybe a question for randomq, or Tam should he ever resurface, is why there is a need to automate the generation of more words when so many end users clearly cannot understand the words already in front of them? Sure, horses to water, same as it ever was, but is there actually value in the mass generation of more text that few, including the prompters, bother to read and even fewer understand? See recent related Punaweb posts for examples.

Without human intelligence what's the point of "AI"?

[HereOnThePrimalEdge]

Without human intelligence what's the point of "AI"?

That’s the key isn’t it?  If it’s a tool in our toolbox we need to read the safety instructions and use it properly.  Otherwise, if given mostly unchecked power with no oversight, no human review, just blind acceptance of the output, or results of a query, what do we get?  Are there any historical or present day examples of what occurs when orders or prompt results are followed with blind obedience and we become yes men without question?  Who benefits from that scenario?

Anyone? Anyone?

[Punatang]

Are there any historical or present day examples of what occurs when orders or prompt results are followed with blind obedience and we become yes men without question? 


You're kidding right?  Can't think of anything recent?

[HiloJulie]

Are there any historical or present day examples of what occurs when orders or prompt results are followed with blind obedience and we become yes men without question?

Measles

[randomq]

We all love Puna (or love to hate it) and we just have to accept that some Punatics are capable of way less reasoning than even the humblest of AIs. At least, with a bit of prompt engineering, we might be able to put them to work for our cause. It certainly seems to have worked for the politicians, who have prompted people into having all kinds of crazy beliefs, acting against their own best interests, etc.

For AI models, once you get pretty far Into a negative context (arguing facts, pushing back against long held beliefs) it's almost impossible to get them to change their tune. But if you start a new context, and focus on something that doesn't trigger them, and build a history of positive interactions, they will actually handle differing opinions more graciously, and can even be convinced to change their opinion.

I think we are all more "promptable" than we'd like to believe.

[ironyak]

I don't think people's susceptibility to "prompting" is in question. From any Sunday sermon of the hundred local variations, to daily HIDOE classroom lectures, to enhanced Kehena drum circles, to reading Punaweb, to enrolling in the SIF, maybe while on Facebook coordinating against the Hawai'i County's agenda to take your truck - local brains and beliefs are obviously being purposefully prompted. It's the difference in content that matters.

One problem with AI is the scale of misinformation and propaganda floods the zone and, unlike in the days of zines and AM talk radio, the noise quickly surpasses the signal, and narrow channels of thoughtful, evidence and reason based info are quickly eclipsed. Combine this with AI companies purposeful design of their products to be highly deferential to the end user, and we're left with a society of cultists and conspiracists who can't separate fact from fiction, but who demand their "god in the machine said so" beliefs be acted upon. (If you haven't run across this yet as many of us have, you will soon enough).

If you think you can prompt these minds back to reason and reality, you might want to get to it and put that to the test - can even start here ;) QAnon and January 6th is just a taste of what is on the horizon when large groups of of people take their online prompted beliefs to their extreme ends.