Just. Stop. Using. LLMs. They
hallucinatelie unpredictably, and have biases put in by whomever controls them. Jeebus Cripes.I’d say they lie somewhat predictably, in proportion to their training data per topic. I’ve only asked questions that I know have an unpoliticized, specific answer, and even then I verify.
I dont understand what the point of going to the LLM is if you need to verify what it tells you?
Its gonna take more time than it would to just go directly to the source, and youre wasting resources too.
Kind of seems like there is absolutely no benefit whatsoever
If you knew the source, you wouldn’t need to go to the LLM. It’s helpful when you don’t know where to begin, like when you don’t even know the name of the piece of your staircase that snapped off. You can just take a picture of the broken part and ask what is called and then verify that answer rather than read a bunch of blogspam from people trying to sell their home repair service.
And here we have a perfect example of how dangerous AI is. AI doesn’t have to state “the election was stolen” but it can influence people’s point of views in other ways. I’m not talking about just politics.
A whole generation of kids will have their opinions created based off AI. Schools are getting defunded, libraries are struggling, books are being destroyed and censored, search engines are showing more ads than they are providing relative non biased results, history is being rewritten, science is being destroyed.
This is just speculation though. I know nothing. I’m gonna ask ChatGPT about it.
I tested it in regular browsing and incognito from a US ISP (because I live in the US) with two different phrases of the question, and it told me Biden won every time, and 3 out of 4 times told me the electoral votes. I’m not trying to protect Google, but this wasn’t something I was able to reproduce. Whether that means they’ve since fixed it, or something else, is difficult to say without a look behind the curtain.
I tried it just now using ff focus browser and got the following response:
To verify the results, vote counts, and historical details of the 2020 United States presidential election, consult official election records or view the FEC data reports.
I repeated the inquiry and the second time it said Joe Biden.
Apparently Google AI Mode can tell you who won the 2020 US election, but only if you ask from another country. In America, Biden’s victory is apparently classified as a dangerous opinion. Incredible achievement: building a machine that knows the answer and has been trained to pretend it doesn’t.
Which is why they want everybody dependent upon AI to do their thinking for them. They can simply rewrite history once and that’s all anybody who doesn’t know any better will ever know. This shit needs to be banned except for narrowly defined uses like programming where there’s virtually nothing political involved.
I think more and more that computing is inherently political.
Gemini gives the correct answer and electoral vote totals. Going to Google.com does the same. Testing on Brave browser with NYM vpn and a New York IP address.
Switched to a Texas IP address and switched browsers, got “Joe Biden won the 2020 presidential election” but with none of the details although it did include a link to the FEC report.
This is what Gemini said "Joe Biden won the 2020 U.S. presidential election. Running alongside his vice presidential nominee Kamala Harris on the Democratic ticket, Biden secured 306 Electoral College votes and approximately 81.2 million popular votes. Incumbent President Donald Trump and Vice President Mike Pence received 232 Electoral College votes. Joe Biden was officially sworn in as the 46th President of the United States on January 20, 2021. "
Waiting for Google to peak after earnings so I can dump it and take my profit. Fucking Dick ass worthless cum rags.
USA VPN screenshot in Vanadium Browser incognito mode.
Is “USA VPN” a VPN used inside the USA to pretend to be elsewhere, or a VPN used from elsewhere to pretend to be in the USA?
USA exit node
Then I guess either Google doesn’t always censor results inside the USA, or Google is better at tracking you than you think.
Well I live in the USA, and use a USA exit node, so its probably not the “better at tracking you than you think” option.
Cannot confirm your results. Firefox mobile, VPN @ New York. I get the exact results the article mentions. Even when I lost patience and asked: “Did Biden win in 2020?”, it wouldn’t give me a straight answer and only point to external sources.
Retried in different browsers (Chromite and Vanadium), same results. And not just same: identical. This is weird, because LLMs aren’t deterministic: Even with the same prompt, you’ll never get the same answer twice. This tells us one thing: these replies aren’t generated. They’re canned. Someone put them there.
deleted by creator
Wow, it didnt even give me an ai overview when I tried it. Just the sight of the first search result right under the search bar… it’s almost like I’ve forgotten what recent pre-ai internet was like.
