Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

National Archives Pushes Google Gemini AI on Employees

404media.co

Post reply on HN
Print view View original

11–15 of 15 posts

Re: National Archives Pushes Google Gemini AI on Employees

#11

Post by verve_rat »

sadeshmukh wrote: post #8
whatsuphotdog wrote: post #5

Earlier quoted context omitted.

Their point is why would would anyone involved in this presentation choose to dub this AI an "expert". They're addressing the overall dishonesty in labeling it as anything other than what it is. They rightfully do not want users interacting with an AI that sports a label that misinforms them about what they're talking to. And that is a very valid concern. One of the chief issues with all of these chat bots is that th…

Nobody is calling it an expert. It's a system prompt to make it perform better. This kind of deliberate misinformation it implicitly promotes with the employee quotes is exactly what I'm talking about.

The people running the demo called it an expert.

Which just demonstrates the complete lack of respect for the domain they are trampling on.

Re: National Archives Pushes Google Gemini AI on Employees

#12

Post by lm28469 »

sadeshmukh wrote: post #4

> “Why is the Generative AI calling itself an ‘expert archivist?’” an employee asked. “It’s called ‘expert archivist’ because that is the prompt we gave it,” someone involved with the demo said. > “I have a serious problem with the ‘expert archivist’ title,” another employee said. > “Same here. If we have a disclaimer saying the generative AI can make things up and yet call it an expert archivist on the same tier as…

Technology is supposed to serve us, if your users tell you you're fucking up you are indeed fucking up

We're talking about actual experts vs some bullshit AI salesman who probably don't know jack shit about the actual job of archiving.

> It's a system prompt for heaven's sake.

That's the whole point, they're archivists not some self proclaimed "pRoMpT EnGiNeErS" tech bros

Re: National Archives Pushes Google Gemini AI on Employees

#13

Post by nine_zeros »

whatsuphotdog wrote: post #6

My partner is an archivist, they attend these sorts of conferences, and we've talked about this frequently. Archivists are at best apprehensive about AI, especially shoddily implemented versions like will almost certainly be the case here. Accuracy is everything, and the way information in the archive is presented to visitors/researchers is paramount . Anything that moves that discussion around that presentation away…

> What's the rush? Why would we settle on 90% accuracy today when we could do it right and get 100%? It's not like this is mission critical information. It's an archive. Take the time and do it right.

Because someone will get a promotion for leading an "initiative".

Re: National Archives Pushes Google Gemini AI on Employees

#14

Post by mistrial9 »

many angles on this current news items -- one that is worth mentioning first, echoes what others have said here.. these professional archivists are the "most expert" at a certain task at a National scale! yet the words "expert archivist" plus black-box corporate run, remote software is on full display, very likely heading towards "doing the job" of these most trained people. And to add insult to injury, in a certain task the success rate of 90% is declared "good enough" when these are people with years of training to never, at all, perform at that low of a level.

Draw your own conclusions about the professionalism and goals of management, emboldened and empowered by "scale" ..

As a counterpoint however.. vast and rapidly expanding knowledgebase is a unique and pressing concern of the modern age. Even world-class, PhD laden and disciplined groups are not keeping up with ingestion and use of the amounts of data emerging. The cure adds to the ailment here, though.. AI itself in current incarnations produces mountains of gibberish or worse, half-truths and half- very plausible and confident nonsense.. This is an important story.

Re: National Archives Pushes Google Gemini AI on Employees

#15

Post by diggernet »

> The specific disclosure will say: “Accuracy: AI-generated summaries and results may not reflect the opinion of NARA and are not guaranteed to be accurate. Historical records often contain factual errors or offensive language, which ArchieAI may repeat or use.”

So they disclose that ArchieAI might repeat factual errors in the records, but don't bother mentioning that ArchieAI might make up fictional history?

Post reply on HN