Bernie talks to AI agent Claude about AI data collection and privacy rights

[Gaugamela]

I spoke to Anthropic’s AI agent Claude about AI collecting massive amounts of personal data and how that information is being used to violate our privacy rights.

What an AI agent says about the dangers of AI is shocking and should wake us up.

[Uncle Joe]

1. Brilliant!

Thanks for the thread Gaugamela

[xocetaceans]

4. One thing that ought to be considered is whether it is useful to go along with the illusion that 'AI' is actually . . .

. . . intelligent in any manner. On what basis could the 'AI' justify any of its responses or extrapolate from them or deduce anything from them?

As is often said with respect to fascism, "Don't surrender in advance." The same is true of 'AI'. One probably should not grant them acceptance so easily: if such acceptance is merited, sure at that point it could be granted, but for now, it is surrendering in advance to set up and publish such an 'interview' scenario.

Most everything that is contained in the early ( up to about 5 minutes into the video when 'AI' model training is mentioned ) response Claude [sic: 'Clod'] makes is a rehashing of the complaints against social media in the last 12 years but with the term "AI" put in place of "algorithm-based social media".

Note how laudatory the responses are. That is corrosive to objectivity and does not serve any sense of any truth that might exist.

Also, note how little is said of the fact that these models are based purely on the appropriation of huge amounts of copyrighted material. The underlying data that supports the statistics these models use is essentially stolen.

In what sense does the model have any 'belief' in the correctness of what Sen. Sanders says late in the video ( see 7:55 in the video, Claude: "You're absolutely right, Senator." )? And when the 'AI' says "I was being naive.", what exactly does that mean? Is that indicative of a mental state which has now been altered by the course of the discussion and has propagated across the entire 'AI' model? Not at all. Ask the 'AI' the same question but argue for a different result: will the 'AI' pliably shift to support than other answer? Most likely.

If you have never tried it, go check how these models do translations from one language to another, but check to see if the translations form a closed loop ( i.e., they should not be altered when going from one language to another and then back to the original language, but they often times do exactly that ). Google's translate seems to have been particularly bad ( often cycling a translation into nonsense ) the last time I checked.

At the end, asking what the 'AI' 'thinks' is just more surrendering in advance to the notion that these tools are in anyway fit to be considered analogous to people.

So, I would just propose that presenting such a video as any sort of authentic exchange is very dangerous. People who literally do not understand anything about technology might very well be taken in by the illusion of that these 'AI' models represent sentience in a box. Nothing is further from the truth: they have no embodiment or, more literally, no skin in the game.

Usually, I would agree with Sen. Sanders's argued positions, but what end does such a video as this one serve? I would argue as above that this acceptance ( even if mock acceptance ) is dangerous.

[highplainsdem]

7. Yes. I'm afraid Bernie just did some free advertising for the genAI companies. And now we're likely

to see Republican politicians or Trump officials post video of Claude agreeing with them.

[usonian]

2. Transcript

https://pastebin.com/raw/3tFr2GQd

[highplainsdem]

3. This is a chatbot whose dataset would include a lot of info, and it's doing the usual chatbot thing and

agreeing with Bernie.

As one of the commenters on YouTube said:

Good stuff. Keeping in mind, of course, that Claude has a detailed profile also of Senator Sanders. Generative AI doesn't "think," as such; it produces plausible responses, and those responses are tailored to the user. There's at least some chance the system was telling him what it thought he'd want to hear; it'd be interesting to see a split-screen of Sanders and, say, Senator Rand Paul asking Claude more-or-less the same questions and see how well the answers match.

Claude is not a person with beliefs and a personality.

Bernie's talking to a machine with no real awareness of what it's saying.

Claude is not imparting wisdom. It's reflecting certain views.

I've been on X at times when a lot of people were asking Grok about the same news story, whether it was true, and Grok's answers showed a wide variation in replies it gave to different people only minutes or even seconds apart (X was set up then to show off what Grok said in response to questions).

There is no real consciousness in these LLMs, no real intelligence. GenAI is sometimes called a fancy autocomplete.

[xocetaceans]

6. I would tend to agree with your post on this topic . . .

. . . and have that statement upthread in a longer form.

[highplainsdem]

8. Just read and recced that, and replied.

[xocetaceans]

12. Thanks. It's unreal that so many people seem enamored with these chatbots and seem so willing to post 'AI' slop here. nt

[WhiskeyGrinder]

5. how embarrassing for him

[highplainsdem]

9. Yes. Extremely.

[Warren_Pointe]

10. If Claude is so intelligent,

why is his logo a drawing of an a$$hole by Kurt Vonnegut?
Expand your bytes, man!

[Avalon Sparks]

11. Love this man

Go Bernie!