Taylor Swift Combats Election Misinformation, Apple Debuts AI iPhones, Oprah Talks Up AI [CNET]

Get up to speed on the rapidly evolving world of AI with our roundup of the week’s developments.

Taylor Swift’s endorsement last week of Kamala Harris for president might be best remembered for the photo of the megastar recording artist with her cat Benjamin Button and her sign-off as a “Childless Cat Lady,” a direct jab at some misogynist Republican messaging. 

But what I found notable was that Swift, the victim of vile pornographic deepfakes earlier this year, said she decided to publicly declare her support for the Democratic nominee after former President Donald Trump used different AI deepfakes of Swift on his social media network to falsely suggest the singer would vote for him.   

“Recently I was made aware that AI of ‘me’ falsely endorsing Donald Trump’s presidential run was posted to his site. It really conjured up my fears around AI, and the dangers of spreading misinformation,” Swift wrote in an Instagram post liked by more than 10 million people. “It brought me to the conclusion that I need to be very transparent about my actual plans for this election as a voter. The simplest way to combat misinformation is with the truth.”

Repeat: The simplest way to combat misinformation is with the truth.

While Swift was able to set the record straight because of her celebrity and powerful platform — she has more than 284 million followers on Instagram and a devoted global fanbase known as Swifties — the post is also a reminder about a lack of federal legislation that would ban or regulate AI deepfakes in the US, according to Gili Vidan, an assistant professor of information science at Cornell University.  

In an email, Vidan said Swift used her massive platform to not only throw her support behind Harris, but also to “highlight the ways she experienced first-hand how new generative AI technologies can be used to undermine” even a world-famous public figure’s stance.

“As was the case when her image was previously manipulated to create abusive synthetic imagery,” Vidan added, “Swift’s ability to effectively use her platform and resources to counter such manipulation serves as a troublesome reminder that many victims do not have similar recourse available to them when trying to reclaim their voices, narratives, and likeness — not legally and not by appealing to social media platforms to remove such content.”

Deepfakes don’t just affect celebrities and the pope. Average people, including young people who’ve been the target of pornographic deepfakes, have almost no recourse when it comes to having such content removed from the internet. The US government has an interesting timeline of the evolution of deepfakes and how they’ve gotten more sophisticated over just the past few years (see page seven here).

There are federal and state efforts to combat AI-generated misinformation (the Harvard Business Review has memorably described such AI content as “botshit“). In July, the Senate passed the Defiance Act of 2024, which provides federal protections to victims of nonconsensual deepfake pornography. That legislation now needs to be approved by the House of Representatives. 

And the State University of New York at Buffalo says it’s developed a free web-based platform for analyzing AI-generated content so you can suss out whether a photo, a video or an audio clip is a deepfake. The tool is called the DeepFake-o-Meter. 

By the way, Swift also used the post to encourage US citizens to sign up to vote ahead of the November election and to do their own research about the candidates. More than 400,000 people have clicked on her link to Vote.gov, up from the average 30,000 visitors that the US government’s voting information website gets on a typical day, a spokesperson for the General Services Administration told USA Today.

The 2020 presidential election was deemed the “most secure” presidential election in US history by Trump administration and other election officials. The 2024 presidential election will be held on Tuesday, Nov. 5. Experts say AI is already being used to generate misinformation about it.

Here are the other doings in AI worth your attention.

Apple adds ‘Apple Intelligence’ to iPhone, but patience required

Apple unveiled four new models of its top-selling iPhone last week, and said the new iPhone 16 will offer generative AI functionality as part of a set of features it announced in June called Apple Intelligence. The AI stuff will also work on last year’s iPhone 15 Pro and iPhone 15 Pro Max, but owners of models older than that are out of luck, because those smartphones don’t have sufficient computational power to run Apple’s AI tech. 

Also, initial Apple Intelligence features will be released in “beta” in a free software update next month (in iOS 18.1), with other features to “roll out later this year and in the months following,” the company said.

CNET reviewers Lisa Eadiccio and Patrick Holland have the details on how AI will work on the iPhone. Two big takeaways for users who can run Apple Intelligence: Apple is putting its focus on integrated AI — features that work together based on users’ needs — and privacy.

“Understanding personal context when delivering answers and carrying out tasks is a big part of Apple’s approach with Apple Intelligence,” said Eadicicco. “Apple seems to be using this tactic as a way to distinguish its own AI efforts from those previously announced by competitors. As an example, the company explained how Apple Intelligence can understand multiple factors like traffic, your schedule and your contacts to help you understand whether you can make it to an event on time.”

As for privacy, the company said last week (and at its Worldwide Developers Conference in June) that its AI engines run on-device. Eadicicco said that’s “generally considered to be more private since information doesn’t have to travel over the internet. At times when a task may require larger models, Apple will use a system called Private Cloud Compute. When needed, tasks that require more computing power will run on servers Apple has specifically created that run on Apple Silicon. Apple Intelligence will decide whether a task can be processed on-device.”

You can find more details on Apple’s AI privacy efforts in this recap I put together.  

OpenAI’s new ‘reasoning’ chatbot thinks before it speaks 

OpenAI, the startup behind ChatGPT, has a new series of large language models, or LLMs, that’ve been trained to “reason through complex tasks and solve harder problems than previous models in science, coding and math,” the company said.

What does that mean? The models, called OpenAI o1, are “designed to spend more time thinking before they respond,” OpenAI said in a blog post.

“With previous models like ChatGPT, you ask them a question and they immediately start responding,” Jakub Pachocki, OpenAI’s chief scientist, told The New York Times. “This model can take its time. It can think through the problem — in English — and try to break it down and look for angles in an effort to provide the best answer.”

A preview version of o1 is now available, and it has limitations. It “isn’t connected to the web, can’t be used with file uploads and has a multitude of API limitations for developers,” reported CNET’s Imad Khan.

“The model doesn’t immediately begin spitting out responses and can take from 10 to 20 seconds to put together a thoughtful answer,” Khan said. “The o1 model, which has also been referred to as ‘Strawberry’ by onlookers (a possible reference to the viral trend of influencers asking AIs to answer how many ‘Rs’ are in the word ‘strawberry’), removes the need for ‘chain-of-thought prompting,’ where users have to ask extra questions of an AI to see its intermediate reasoning. Instead, the model is designed to show its reasoning by default.”

What’s the takeaway for you? Well, we’re living through the evolution of gen AI, so consider this a step in that development. Instead of “hallucinating” (that is, giving you an answer that sounds right but is actually wrong) or choking on simple math problems, the o1 chatbot is intended to think about the answer to your question, step by step, before it gives you an answer — moving it a step closer to the way humans solve problems. 

In related news, according to tech insider site The Information, OpenAI Chief Operating Officer Brad Lightcap told staff that OpenAI has more than 10 million paying subscribers and another 1 million subscribers who’ve bought higher-priced business licenses. That’s just a reminder that OpenAI reportedly continues to lose money and needs to show that it’s got a chance of making money, if it wants to go public in the near future.

Oprah talks AI in TV special to separate reality, hype — we hope 

Oprah Winfrey spoke with Microsoft co-founder Bill Gates, OpenAI CEO Sam Altman and tech reviewer Marques Brownlee in an hour-long special on ABC last week called AI and the Future of Us. 

“The reason why I wanted to do this special is because I’m right where most of you all are with AI,” Winfrey said in an Instagram Reels post. “You hear it all the time and you hear it’s going to change our lives, and we are aware of it on our phones and you’re aware of it in your cars and you’re aware of it for certain areas of your life, but what is really coming? What is happening with AI?”

In addition to people working in the AI industry or around it, Winfrey also spoke to FBI Director Christopher Wray, Pulitzer Prize-winning author Marilynne Robinson and the co-founders of the Center for Humane Technology, which focuses on the “negative effects” of technology.

But before it aired, the show was already being called out for who wasn’t interviewed, with critics noting that Winfrey talked with more AI boosters than people who have serious concerns about how gen AI is being developed and shared, as CNET noted.

In an X post on Aug. 30, Los Angeles Times tech columnist and Blood in the Machine newsletter writer Brian Merchant called the special an “extended sales pitch for the generative AI industry at the moment when its fortunes are flagging and the AI bubble is threatening to burst.”

Hey, Oprah — happy to give you a list of people who might provide a more balanced view of AI, if you’re planning a follow-up show.

For her part, Winfrey, in an interview with a Washington Post columnist who helped her produce the special, said she tried not to “scaremonger people, but to have a healthy amount of suspicion and concern about the negative outcomes.” 

She added: “There’s no question that there are positives here. But if you have a criminal mind and are the kind of person who is thinking about how you can get over and how you can scam people, [AI] is just an open door to that. Which means the rest of us have to develop a suspicion muscle. And that’s going to change the way we interact with everything that we read and see and hear. It’s going to change how we interact with each other. We have to ask: Is this real? Where is it coming from? I just know that that’s no way to live. I mean, there’s no peace in suspicion. There’s no peace in constant suspicion.”

Winfrey also said she was “surprised” by OpenAI CEO Altman’s “clarity and efficacy and being able to speak about [AI] in a language that humans can understand. I didn’t expect that.” 

I haven’t watched the show as of this writing, but I’m happy to give you my take next time. I already know I won’t be surprised by a lot of things coming out of the mouths of AI boosters. 

A16z shares its list of the Top 100 gen AI consumer apps

Speaking of AI boosters, venture capital firm Andreesen Horowitz, also known as A16z, released its third compilation of the Top 100 gen AI Consumer Apps. Even taking into account that the firm is an AI investor that’s poured at least $1 billion into AI startups, I still think the lists — broken down by monthly active visits and monthly active users — are worth a look.

The top five gen AI tools by visits are ChatGPT, Character.ai, Perplexity.ai, Anthropic’s Claude, and song-maker Suno. Though ChatGPT continues to lead the pack among general chatbots or assistants, Perplexity and Claude have also gained traction, A16z’s data shows.

The last time A16z compiled the list was in March, and it said that this time around, compared with that last effort, “nearly 30% of the companies were new.” Newcomers include Lumai.AI, PicWish, AI Novelist and some AI tools created by Bytedance, TikTok’s parent company.

The firm offers this insight: “Fifty-two percent of the companies on the web list are focused on content generation or editing, across modalities — image, video, music, speech, and more. Of the 12 new entrants, 58% are in the creative tool space.” 

Scientists use voice recordings to spot chronic high blood pressure

Klick Applied Sciences, which has developed AI tools that can detect Type 2 diabetes using 6-to-20-second smartphone recordings, has developed a new tool that uses a person’s voice and AI to detect chronic high blood pressure. 

In a paper published by IEEE Access, researchers explained why this could be a game changer for the 25% of the global population affected by hypertension, which the World Health Organization has called the “silent killer.” Half those people are unaware of their condition, and over 75% live in low- or middle-income countries, Klick said, citing WHO data. 

“Conventional methods of measuring blood pressure (and, accordingly, identifying hypertension) include using an arm cuff (sphygmomanometry) or an automatic blood pressure measurement device. However, these methods may require technical expertise, specialized equipment, and may not be readily accessible to people in underserved areas,” the company wrote in its press release.   

As part of its study, Klick said, it asked 245 people to “record their voices up to six times daily for two weeks” using a mobile app it developed. Company scientists found that its model detected high-blood pressure 84% of the time for women and 77% of the time for men. 

Like its diabetes solution, the hypertension app analyzes “hundreds of vocal biomarkers that are indiscernible to the human ear, including the variability in pitch (fundamental frequency), the patterns in speech energy distribution (Mel-frequency cepstral coefficients), and the sharpness of sound changes (spectral contrast).”

Super interesting, right? Considering that over half of the world’s population now has a smartphone. WHO says that “in many places, people are more likely to have access to a mobile telephone than to clean water.” 

“Voice technology has the potential to exponentially transform healthcare, making it more accessible and affordable, especially for large, underserved populations,” said Jaycee Kaufman, Klick Labs research scientist and co-author of the study. “Our ongoing research increasingly demonstrates the significant promise of vocal biomarkers in detecting hypertension, diabetes, and a growing list of other health conditions.”

AI may spur renewed interest in more-stable blue-collar jobs 

With tech and other companies cutting staff as they redirect their money into AI initiatives, it makes sense that some workers might consider turning to “stable” blue-collar jobs and invest in non-AI skills as older workers retire from those skilled trade positions.

That’s the take from data compiled by ResumeCoach, which said that over 1.7 million positions spanning seven blue-collar jobs are expected to be created over the next 10 years. That includes 1 million job openings in 2024, in sectors such as construction and manufacturing.

ResumeCoach also found that green investments may prompt a “green collar boom,” with jobs like wind turbine service technicians set to grow by 45%.

As for the top seven job categories for blue-collar workers in 2024-2025, ResumeCoach said there are opportunities in construction; production; transportation; food preparation; farming, fishing and forestry; installation, maintenance and repair; and building and ground cleaning and maintenance.  

In the next decade, ResumeCoach said, “traditional blue-collar jobs, such as electricians and plumbers, aren’t going anywhere anytime soon. Roles like personal care aides, restaurant staff, and truck drivers are among the top 20 occupations with projected numeric change in employment.”

Though some of these occupations might also be affected by AI — making it a good idea to learn new skills in whatever job you take on — ResumeCoach added that workers who prefer the “hands-on, practical nature of blue-collar work” may find these careers both “fulfilling and resilient in a changing job market.”

Food for thought.