What’s scariest is that you’d think it unfathomable that a company such as Google would make such a mistake, but considering the sheer speed at which they are shipping their products, mistakes like this will continue taking place.
While I'm very surprised the model generated historical white people as non-white; that signals to me that, once again, these models are much dumber and unconscious about the real world that many people thought them to be.
What happens when #Artificial Intelligence is autonomously controlling more and more of our societal livelihood and something like this happens?  

https://blog.google/products/gemini/gemini-image-generation-issue/

Gemini image generation got it wrong. We'll do better.

Artificial Intelligence

What’s scariest is that you’d think it unfathomable that a company such as Google would make such a mistake, but considering the sheer speed at which they are shipping their products, mistakes like this will continue taking place. While I'm very surprised the model generated historical white people as non-white; that signals to me that, once again, these models are much dumber and unconscious about the real world that many people thought them to be. What happens when #Artificial Intelligence is autonomously controlling more and more of our societal livelihood and something like this happens?    https://blog.google/products/gemini/gemini-image-generation-issue/

Jerome Cody

the dangers of this type of thing happening increase by the day....

Microsoft’s Beijing-based research group published a new open-source AI model on Tuesday, only to remove it from the internet hours later after the company realized the model hadn’t gone through adequate safety testing

https://www.theinformation.com/briefings/microsoft-takes-down-wizardlm-ai-model-china-safety-toxicity-checks?utm_campaign=article_email&utm_content=article-12674&utm_medium=email&utm_source=sg&rc=gzsl71

the dangers of this type of thing happening increase by the day....  Microsoft’s Beijing-based research group published a new open-source AI model on Tuesday, only to remove it from the internet hours later after the company realized the model hadn’t gone through adequate safety testing  https://www.theinformation.com/briefings/microsoft-takes-down-wizardlm-ai-model-china-safety-toxicity-checks?utm_campaign=article_email&utm_content=article-12674&utm_medium=email&utm_source=sg&rc=gzsl71

Bobby

Meta's AI image creator is facing criticism for its inability to generate people from diverse racial backgrounds, including mixed-race couples. The Verge was the first to report that the Imagine AI system consistently failed to generate images of White and Asian couples together, even when explicitly prompted to do so.

https://www.cnn.com/2024/04/04/tech/meta-ai-image-generator-interracial-couples/index.html

Meta's AI image creator is facing criticism for its inability to generate people from diverse racial backgrounds, including mixed-race couples. The Verge was the first to report that the Imagine AI system consistently failed to generate images of White and Asian couples together, even when explicitly prompted to do so.  https://www.cnn.com/2024/04/04/tech/meta-ai-image-generator-interracial-couples/index.html

Meta’s AI image generator is coming under fire for its apparent struggles to create images of couples or friends from different racial backgrounds.

Meta’s AI image generator really struggles with the concept of interracial couples | CNN Business

Semafor found that Adobe's AI image creation tool generated many of the same historically inaccurate images as Google's Gemini, including Black Nazi soldiers and female Founding Fathers.

https://www.semafor.com/article/03/13/2024/adobe-firefly-repeats-the-same-ai-blunders-as-google-gemini

Semafor found that Adobe's AI image creation tool generated many of the same historically inaccurate images as Google's Gemini, including Black Nazi soldiers and female Founding Fathers.  https://www.semafor.com/article/03/13/2024/adobe-firefly-repeats-the-same-ai-blunders-as-google-gemini

The problems with Adobe’s image creation tool illustrate the challenges tech companies face across the industry.

Adobe Firefly repeats the same AI blunders as Google Gemini | Semafor

There was a moment last summer when Elon Musk had challenged Mark Zuckerberg to a mixed martial arts fight, and I, for once, had sympathy for Meta Platforms’ public relations team. The owner of X was proposing he settle his differences with Zuckerberg by fighting him in the Roman Colosseum. 

It was absurd. And yet, months later, it seems to be a metaphor for where more and more companies find themselves with Musk: facing unpredictable guerrilla attacks from a man with 175 million online followers, many of whom support his every keystroke.

In the months since Musk’s Zuck throwdown (which ended with a whimper, not an arm bar, from Musk, who backed out because of an injury), Musk has aimed a figurative bazooka at a greater variety of business foes. 

After Disney pulled its advertising from X over Musk’s support for an antisemitic conspiracy theory, he urged people to boycott the company and publicly told CEO Bob Iger to “fuck yourself.” 

He is waging a full-throated attack against Google’s Gemini artificial intelligence, which is under fire for producing ahistorical images to make them more diverse. “Google Gemini is super racist & sexist!” he posted on X in an item that got more than 100,000 views. He’s repeatedly supported and amplified the opinions of people who feel the same way.

And he sued OpenAI, which he helped found back when it was just a nonprofit, for breaching its agreement to benevolently develop AI when it became a business tied up with Microsoft.

I think these attacks are working. On AI bias, Meta’s AI image generator was making some of the same decisions as Google’s, but fewer people are paying attention to Meta’s, perhaps because its image generator gets less use. But it seems likely to me that the relentless posting Musk and friends like Marc Andreessen are doing on the Google issue must have something to do with it. 

So what’s a company under Musk assault to do? 

The simple answer is: You fire up your base. You will never convince Elon followers you are right. But you must get your message to your customers and employees loud and clear. “You have to jiujitsu the criticism to advance your objective,” Elliot Schrage, who was a longtime comms and policy chief at both Google and Meta, told me.  

In some ways that is what Zuckerberg did when he took on Musk over MMA. He said he would fight Musk in a sport he loves and trains in if Musk were serious about it, and he responded to Musk’s taunts with phrases like “Send Me Location,” which is apparently the catchphrase of a famous fighter. MMA fans on social media ate it up.

OpenAI’s leaders were also appealing to the company’s base when they responded to Musk’s lawsuit on its website with the following: “We’re sad that it’s come to this with someone whom we’ve deeply admired—someone who inspired us to aim higher, then told us we would fail, started a competitor, and then sued us when we started making meaningful progress towards OpenAI’s mission without him.”

It is a classic Sam Altman zinger aimed right at the engineers and young tech talent OpenAI is trying to recruit. “We are on to something big here and are beating the odds” is the subtext of this message. “Come join us.” 

Google should try something similar. Despite major internal problems, including around how it builds and launches products, the company still harbors probably the world’s largest group of AI talent, computing power and distribution. Alphabet CEO Sundar Pichai and CEO of Google DeepMind Demis Hassabis can promise their team far more access to training hardware and capital than virtually anyone. And they can affect more people on the planet than virtually any other business. 

Musk is picking on them because they are powerful and important. They should fix their mistakes and apologize for them, but Google’s leaders should also find a way to talk about their advantages, too.

There was a moment last summer when Elon Musk had challenged Mark Zuckerberg to a mixed martial arts fight, and I, for once, had sympathy for Meta Platforms’ public relations team. The owner of X was proposing he settle his differences with Zuckerberg by fighting him in the Roman Colosseum.   It was absurd. And yet, months later, it seems to be a metaphor for where more and more companies find themselves with Musk: facing unpredictable guerrilla attacks from a man with 175 million online followers, many of whom support his every keystroke.  In the months since Musk’s Zuck throwdown (which ended with a whimper, not an arm bar, from Musk, who backed out because of an injury), Musk has aimed a figurative bazooka at a greater variety of business foes.   After Disney pulled its advertising from X over Musk’s support for an antisemitic conspiracy theory, he urged people to boycott the company and publicly told CEO Bob Iger to “fuck yourself.”   He is waging a full-throated attack against Google’s Gemini artificial intelligence, which is under fire for producing ahistorical images to make them more diverse. “Google Gemini is super racist & sexist!” he posted on X in an item that got more than 100,000 views. He’s repeatedly supported and amplified the opinions of people who feel the same way.  And he sued OpenAI, which he helped found back when it was just a nonprofit, for breaching its agreement to benevolently develop AI when it became a business tied up with Microsoft.  I think these attacks are working. On AI bias, Meta’s AI image generator was making some of the same decisions as Google’s, but fewer people are paying attention to Meta’s, perhaps because its image generator gets less use. But it seems likely to me that the relentless posting Musk and friends like Marc Andreessen are doing on the Google issue must have something to do with it.   So what’s a company under Musk assault to do?   The simple answer is: You fire up your base. You will never convince Elon followers you are right. But you must get your message to your customers and employees loud and clear. “You have to jiujitsu the criticism to advance your objective,” Elliot Schrage, who was a longtime comms and policy chief at both Google and Meta, told me.    In some ways that is what Zuckerberg did when he took on Musk over MMA. He said he would fight Musk in a sport he loves and trains in if Musk were serious about it, and he responded to Musk’s taunts with phrases like “Send Me Location,” which is apparently the catchphrase of a famous fighter. MMA fans on social media ate it up.  OpenAI’s leaders were also appealing to the company’s base when they responded to Musk’s lawsuit on its website with the following: “We’re sad that it’s come to this with someone whom we’ve deeply admired—someone who inspired us to aim higher, then told us we would fail, started a competitor, and then sued us when we started making meaningful progress towards OpenAI’s mission without him.”  It is a classic Sam Altman zinger aimed right at the engineers and young tech talent OpenAI is trying to recruit. “We are on to something big here and are beating the odds” is the subtext of this message. “Come join us.”   Google should try something similar. Despite major internal problems, including around how it builds and launches products, the company still harbors probably the world’s largest group of AI talent, computing power and distribution. Alphabet CEO Sundar Pichai and CEO of Google DeepMind Demis Hassabis can promise their team far more access to training hardware and capital than virtually anyone. And they can affect more people on the planet than virtually any other business.   Musk is picking on them because they are powerful and important. They should fix their mistakes and apologize for them, but Google’s leaders should also find a way to talk about their advantages, too.

Sam Mendoza

Not surprising seeing the speed at which Google is shipping its AI products! from 1.0 to 1.5 they took only 2 months, which is very, very short in AI terms.

Jerome Cody

A group of likeminded people in Artificial Intelligence are talking about this.