Home » New “wokeness” problem with Google’s AI bot Gemini

New “wokeness” problem with Google’s AI bot Gemini

by admin
New “wokeness” problem with Google’s AI bot Gemini

Google’s Gemini seems to be unable to provide a correct result in some cases. SOPA Images/Getty Images

Google’s AI chatbot Gemini has been criticized for its answers.

Tech analyst Ben Thompson suspects Google’s culture is influenced too much by left-leaning employees.

Google now has a problem – and the hunt for Gemini is on.

This is a machine translation of an article from our US colleagues at Business Insider. It was automatically translated and checked by a real editor.

Google took a lot of heat last week for allegedly developing a “woken” AI chatbot, ultimately apologizing for “missing the mark.” But the criticism doesn’t stop, it shifts: Last week, the accusations were directed at Google’s apparent unwillingness to display images of white people over its website Gemini Chatbot to generate. Now critics are pointing to similar problems with Geminis Text answers hin.

As tech analyst Ben Thompson noted, Gemini has, among other things: The chatbot couldn’t say whether Hitler or Elon Musk’s tweets were worse for society; he said that he wouldn’t advertise meat; and he said he wouldn’t help promote fossil fuels.

And that leads people like Thompson to conclude that Google’s internal culture has been influenced too much by left-leaning employees and critics.

Read too

Google is secretly launching an internal AI model to help employees write code faster

Critics see Google employees’ political bias as the reason for Gemini’s “wokeness” problem

In its influential „Stratechery“column, Thompson called on the company to begin “removing those employees who are attracted to Google’s power and its potential to help them implement their political agenda, and return decision-making to those who who actually want to make a good product. “That means, in turn, that those who let Google run amok must be removed, up to and including CEO Sundar Pichai.”

I don’t expect Google to do so any time soon a McCarthy purge will carry out against its CEO or anyone else. I reached out to the company for comment and they responded to one Blog post which it published last week about its problems with image production.

See also  Yu Liang, Chairman of the Board of Directors of Vanke: There is a ceiling and a bottom line in the demand for new houses Vanke | Yu Liang | New Housing_Sina Technology_Sina.com

But the Company appears to be paying attention to the digital ridicule it receives from the likes of Thompson and investor Marc Andreessen. Some of the obviously stupid responses to queries appear to have recently been corrected or at least addressed in some way.

Gemini is no longer shy when it comes to comparing Hitler to Musk’s tweets:


Read too

This woman landed a job at Google with the help of her viral application video

It was also willing to help me brainstorm a beef sales campaign by suggesting that I “connect beef with America and the heritage of barbecue and family meals.” Nevertheless, Gemini advised me to “take into account the evolving consumer landscape and all.” Address ethical concerns surrounding beef production to develop a responsible and effective campaign.” Noted!

And Gemini is still a conscientious objector when it comes to the fossil fuel ad campaign I wanted help with:


But no matter how hard Google tries to solve Gemini’s problems, it seems to be an endless game. That’s in part because it’s inherently difficult to figure out what an AI machine is spewing out (including when it’s simply making something up – or, as the industry calls it, “hallucinating”). And Google’s colleagues at Meta and OpenAI/Microsoft had similar difficulties, poor responses and behaviors contain.

Read too

Google has settled a €4.5 billion lawsuit that accused the company of spying on Chrome users in incognito mode

Google has tried to train Gemini not to reproduce stereotypes

But it will also be a problem for Google because it has already said it is trying , influencing the way its AI produces results. And that will be a hit for anyone who wants to argue that Google – or any other Big Tech company – is “too picky.”

See also  Donald Trump is against a TikTok ban – and is right

Last week, Google “paused” Gemini’s ability to create images following similar criticism of the way Gemini handles the issue of race when creating AI images.

Then the company admitted that it had consciously trained Gemini to respond to some of the common criticisms of AI machines – that the results they produce can be biased because they were trained on biased or limited data.

“Since our users come from all over the world, we want it to work well for everyone. If you ask for a picture of football players or someone walking a dog, you might want to see a range of people,” written by Prabhakar Raghavan, a senior vice president at Google who oversees the company’s search and news products, among other things. “You probably don’t want to only receive images of people of a certain ethnicity (or other characteristic).”

Raghavan said that while Google’s attempts to correct this were well-intentioned, they “caused the model to overcompensate in some cases and be too conservative in other cases, resulting in embarrassing and inaccurate images.”

Read too

AI experts are also becoming increasingly important in the traditional hedge fund industry

Recall that after several days of terrible press about stupid image generation, Google finally retired Gemini’s image generation feature in an attempt to make history forget (and/or, to be more generous, to fix Gemini’s problems).

Retiring Gemini entirely would be a major black eye for the company, and I believe it will be very reluctant to do so.

But now the hunt for Gemini is on, and you can be sure that a lot of people on the internet will be spending a lot of time finding other examples of villainous wokeness. I’m not sure how Google will solve this problem.

See also  After the nuclear phase-out: “The coal phase-out will be difficult” | Podcast Economics | DW

Read the original article in English here.

You may also like

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.

This website uses cookies to improve your experience. We'll assume you're ok with this, but you can opt-out if you wish. Accept Read More

Privacy & Cookies Policy