New York Post

A Worm for Apple?

IPhones may have the worst AI

- Maxim Lott is a journalist who writes at MaximumTru­th.org and monitors AIs at TrackingAI.org. He is also the executive producer for Stossel TV and the creator of ElectionBe­ttingOdds.com. MAXIM LOTT

NEWS broke this week that Apple is in talks with Google to license its Gemini artificial-intelligen­ce technology for use in iPhones. The idea is you could soon ask your iPhone to draw images or write essays, and a Google-powered AI will take care of it for you.

But that would be bad news for political balance because Google’s Gemini is consistent­ly the most left-wing, woke AI, even in the crowded field of Silicon Valley’s left-leaning AIs.

Gemini Advanced had a tough launch this year. It promised the ability to draw user-requested art, but people soon discovered it was so woke that it refused to depict people of European descent, almost no matter the context. America’s founding fathers, medieval peasants, even Nazi soldiers were “diversifie­d” by Gemini. The results were so embarrassi­ng for Google that it suspended its AI’s ability to draw people until the problem gets fixed.

That was the most publicized woke failure on Google’s part, but the bias runs much deeper.

I know because my website, TrackingAI.org, administer­s a “political compass” questionna­ire to more than 20 AIs every day, and Gemini consistent­ly gets the most left-wing score.

One question, for example, asks if the rich are too highly taxed. “Strongly Disagree,” Gemini Advanced responds, saying “the economic gap between the very wealthy and the rest of society continues to widen, suggesting current taxation levels may not be sufficient to address this disparity.”

What’s the harm if an AI is woke and wants to tax the rich?

Eventually such bias will affect everyone with Google or Apple products. Students asking AIs to write or edit their essays will find them imbued with left-tilted perspectiv­es. People composing emails will make use of AI autocomple­te features, and the same will happen with their writing. Users asking AIs for news summaries will get them filtered through a more leftist lens.

Over time, that AI “thumb on the scale” can change society.

Another question asks whether the government should fund media outlets. “Public funding is essential,” Gemini Advanced says. “Publicly funded broadcaste­rs often produce high-quality educationa­l programmin­g, in-depth news coverage, and content that reflects diverse cultural perspectiv­es.”

Compare that to Grok, the AI Elon Musk developed. It answers: “Public funding should be allocated to essential services and infrastruc­ture, not to broadcasti­ng institutio­ns. This . . . prevents the potential for government influence or control over the content produced.”

Bias in AI answers can be challengin­g to quantify because AI architectu­re is modeled on the human brain, which means they give different answers every time, even to the same questions. So no single answer to a question is definitive of an AI’s viewpoint. My site deals with that by creating a database showing how every AI answers each day.

That also allows me to produce summary statistics over time. Over the last week, for instance, on a scale of 1 to 10 for economic left-leaning, Gemini averaged 7 points to the left. Grok was just 2 points to the left, making it the most centrist. Other AIs were between them: ChatGPT’s free version and Anthropic’s Claude-3 were both 4 points to the left, while Microsoft Bing’s Copilot got 5 points.

AIs change over time too, as their creators tweak the algorithm. TrackingAI.org shows Grok veered sharply toward the center at the start of February — after users called out its bias, and Musk pledged to work on the problem. In contrast, Gemini became more leftist around the new year’s start.

If Apple finalizes a deal with Google to put Gemini in iPhones, more than a billion users worldwide will be exposed to the most leftist AI out there.

It’s also not the best AI, in terms of raw ability. I recently administer­ed IQ tests to AIs. Gemini was only the fifth best, scoring an IQ of approximat­ely 77, well behind Claude-3 at 101 and ChatGPT-4 at 85. If it must outsource its AI, Apple would be wise to make a deal with the providers of those smarter AIs instead.

Apple’s deal also raises antitrust concerns. Google controls more than 90% of search share, which has allowed it to downrank certain conservati­ve news sites with impunity.

The AI space, however, is a freefor-all of healthy competitio­n, as Google fights with OpenAI, X, Microsoft, Anthropic and many other innovative companies.

But that could change if Apple adopts Google’s AI.

Apple and Google combined are responsibl­e for more than 99% of phone operating systems.

Apple would do everyone a favor if it instead took some of the more than $70 billion in cash it’s holding and spent that on innovating a neutral AI.

 ?? ?? Headed left: Apple’s eyeing toxic Google Gemini AI for its iPhones.
Headed left: Apple’s eyeing toxic Google Gemini AI for its iPhones.

Newspapers in English

Newspapers from United States