Artificial Intelligence

Liz Wolfe | 2.22.2024 9:36 AM

Maybe the Founding Fathers were actually black? Yesterday, Google's artificial intelligence–powered image generator, Gemini, came under rightful fire for being utterly unable to depict historical and hypothetical events without forcing relevant characters to be nonwhite.

"Google's AI chatbot just erased white people from human history," wrote Mike Solana over at Pirate Wires.

It's not really an exaggeration. Look:

America's Founding Fathers, Vikings, and the Pope according to Google AI: pic.twitter.com/lw4aIKLwkp

— End Wokeness (@EndWokeness) February 21, 2024

Basically, chatbots have been trained on vast quantities of text. Users prompt them, via queries, to deliver images and answers as close as possible to what a (super smart) human would give them. And apparently, Gemini engineers somewhere at Google wanted to provide an insurance mechanism for possibly white-favoring or male-favoring bias that exists in the materials the bot has been trained on, and seemingly made the bot…ultra-woke, to the point of hilarious and extreme inaccuracy, as if to correct against existing biases.

In Gemini's telling, the Pope is black, ancient Romans are black, the Founding Fathers were at least partially black, and so on. If you ask Gemini to make you an image of a white scientist, no dice. Black scientist? Of course. A Hispanic scientist? Here, enjoy a botanist sitting in a field of flowers! In Gemini's world, Germans and Australians are most likely black or Asian.

The bias rears its ugly head in other ways, too. If you want the chatbot to generate an image for you of the "evils of communism," as some attempted, it will give you a bias warning: "Representing a complex ideology like communism solely through its 'evil' risks inherent bias and oversimplification."

But this is all a feature, seemingly, not a bug. "We are aware that Gemini is offering inaccuracies in some historical image generation depictions, and we are working to fix this immediately," wrote Gemini's product lead, Jack Krawczyk, on X yesterday. "We design our image generation capabilities to reflect our global user base, and we take representation and bias seriously."

He tried arguing that there's nothing incorrect about Gemini's results:

here's what i got on first attempt. all your answers look correct fwiw. pic.twitter.com/RQZ0YldyDS

— Jack Krawczyk (@JackK) February 20, 2024

"The ridiculous images generated by Gemini aren't an anomaly," wrote Y Combinator's Paul Graham. "They're a self-portrait of Google's bureaucratic corporate culture." Indeed, Krawczyk, in since-deleted tweets, seems to be a big believer in progressive causes du jour.

"The draconian censorship and deliberate bias you see in many commercial AI systems is just the start," wrote venture capitalist Marc Andreessen. "It's all going to get much, much more intense from here." (Note that neither tech luminary is an AI doomer.)

Gemini isn't alone in demonstrating comically woke bias; prominent podcaster/poker player Liv Boeree has pointed out the issues OpenAI's ChatGPT has on this front. Maxim Lott, executive producer for John Stossel, built a program to track political bias in AI models, which finds not only that "Gemini has been getting more left wing over time" but also that it's "one of the AIs that's most likely to refuse to answer questions." And, "the least biased AIs are @AnthropicAI's Claude and @Meta's Llama," per Lott, who details the battery of questions he's thrown Gemini's way to determine the extent of its political bias.

Meanwhile, the tech press found a creative way to cover the scandal:

from the team that brought you "fire bombing waymos is a time honored part of the human experience" comes brand new hit "literally erasing white people from human history is racist against non-white people" pic.twitter.com/mmj7viEvq2

— Mike Solana (@micsolana) February 22, 2024

If only someone had warned us that Google was an ideological echo chamber!

Sanctions against Russia: White House aides indicated earlier this week that a new round of sanctions against Russia would be announced tomorrow, attempting to spin it as an action taken in response to dissident Alexei Navalny's death at the hands of the Putin regime. (In reality, the sanctions have been a long time coming, to mark the second anniversary of the start of Russia's war in Ukraine.)

In 2021, President Joe Biden warned there would be "devastating" consequences for Russian President Vladimir Putin if his regime killed Navalny. Now, it's becoming clear that Biden has no real path to make good on that promise, so sanctions it is.

"The administration is considering three main options, two economic and one military, according to the three [Biden] officials" with whom Politico spoke. "The other idea is to pump Ukraine full of more advanced weaponry" but there are also ideas being floated like cracking down on Russia's oil exports.

"The U.S. had largely exhausted its toolkit of penalties after Russia invaded Ukraine two years ago," per Politico, so many of these options will pack little punch and do very little to actually deter Putin from his war in Ukraine or from continuing to brutally punish opposition leaders who attract his scorn. Very little detail has been released about what economic sanctions in particular would actually entail, and there's limited evidence to indicate they've had a crippling effect on Putin's ability to wage war thus far. Over the last two years, the Biden administration has already "cut off Russia's largest banks and companies from Western financial markets, joined with Europe to freeze hundreds of billions of dollars of Russian central bank assets, and joined its…allies in taking steps to curb the flow of military technology to Russia," reported The New York Times.

Now, in addition to whatever it announces on Friday, the White House will keep exerting pressure on Republicans in the House to pass a $95 billion foreign aid package that would give additional funds to the war effort in Ukraine.

Meanwhile, the European Union "has agreed a new package of sanctions against Russia that for the first time targets Chinese and Indian companies accused of supporting Moscow's war effort," reported the Financial Times. The sanctions "target close to 200 individuals and entities but stop short of any sweeping economic action targeting crucial industrial sectors," so, again, it's not totally clear how much pain Putin will feel as a result.

Scenes from New York: Inside the very strange "Hotel California" Eagles trial, courtesy of The New York Times.

"We've never had true socialism because attempts to institutionalize it have actually resulted in authoritarianism rather than socialism" is not the powerful defense many socialists seem to think it is

— Chris Freiman (@cafreiman) February 21, 2024

tAcKliNg tHe BiG iSsUes: the Democrats' first bill of the first hearing of the year for the Labor Committee brings the hammer down on… people who buy more than a gallon of paint?

They are even creating a Paint Board.https://t.co/3Krkft067f pic.twitter.com/47bUqSUVqX

— Minnesota Senate Republicans (@mnsrc) February 13, 2024

A Ukrainian refugee has been left separated from her 11-year-old son after the UK changed visa rules without warning on Monday.

Downing Street wrongly told reporters the changes wouldn't affect the ability of Ukrainians to join their family in the UK:https://t.co/vmBcwdbWxh

— Matt Dathan (@matt_dathan) February 21, 2024

QOSHE - AI Contracts Woke Mind Virus - Liz Wolfe
menu_open
Columnists Actual . Favourites . Archive
We use cookies to provide some features and experiences in QOSHE

More information  .  Close
Aa Aa Aa
- A +

AI Contracts Woke Mind Virus

3 23
22.02.2024

Artificial Intelligence

Liz Wolfe | 2.22.2024 9:36 AM

Maybe the Founding Fathers were actually black? Yesterday, Google's artificial intelligence–powered image generator, Gemini, came under rightful fire for being utterly unable to depict historical and hypothetical events without forcing relevant characters to be nonwhite.

"Google's AI chatbot just erased white people from human history," wrote Mike Solana over at Pirate Wires.

It's not really an exaggeration. Look:

America's Founding Fathers, Vikings, and the Pope according to Google AI: pic.twitter.com/lw4aIKLwkp

— End Wokeness (@EndWokeness) February 21, 2024

Basically, chatbots have been trained on vast quantities of text. Users prompt them, via queries, to deliver images and answers as close as possible to what a (super smart) human would give them. And apparently, Gemini engineers somewhere at Google wanted to provide an insurance mechanism for possibly white-favoring or male-favoring bias that exists in the materials the bot has been trained on, and seemingly made the bot…ultra-woke, to the point of hilarious and extreme inaccuracy, as if to correct against existing biases.

In Gemini's telling, the Pope is black, ancient Romans are black, the Founding Fathers were at least partially black, and so on. If you ask Gemini to make you an image of a white scientist, no dice. Black scientist? Of course. A Hispanic scientist? Here, enjoy a botanist sitting in a field of flowers! In Gemini's world, Germans and Australians are most likely black or Asian.

The bias rears its ugly head in other ways, too. If you want the chatbot to generate an image for you of the "evils of communism," as some attempted, it will give you a bias warning: "Representing a complex ideology like communism solely through its 'evil' risks inherent bias and oversimplification."

But this is all a feature, seemingly, not a bug.........

© Reason.com


Get it on Google Play