In Norway, electric vehicles are selling like hot cakes – and Tesla models the most of all. CC-licensed photo by Norsk Elbilforening on Flickr.
You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.
A selection of 9 links for you. Use them wisely. I’m @charlesarthur on Twitter. On Mastodon: https://newsie.social/@charlesarthur. Observations and links welcome.
OpenAI CEO Sam Altman warns of AI’s potential harm, wants regulations • The Washington Post
Cat Zakrzewski, Cristiano Lima and Will Oremus:
OpenAI chief executive Sam Altman delivered a sobering account of ways artificial intelligence could “cause significant harm to the world” during his first congressional testimony, expressing a willingness to work with nervous lawmakers to address the risks presented by his company’s ChatGPT and other AI tools.
Altman advocated for a number of regulations — including a new government agency charged with creating standards for the field — to address mounting concerns that generative AI could distort reality and create unprecedented safety hazards. The CEO tallied “risky” behaviors presented by technology like ChatGPT, including spreading “one-on-one interactive disinformation” and emotional manipulation. At one point he acknowledged AI could be used to target drone strikes.
“If this technology goes wrong, it can go quite wrong,” Altman said.
Yet in nearly three hours of discussion of potentially catastrophic harms, Altman affirmed that his company will continue to release the technology, despite likely dangers. He argued that rather than being reckless, OpenAI’s “iterative deployment” of AI models gives institutions time to understand potential threats — a strategic move that puts “relatively weak” and “deeply imperfect” technology in the world to understand the associated safety risks.
…Sen. Richard Blumenthal (D-Conn.), who chairs the subcommittee, said Altman’s testimony was a “far cry” from past outings by other top Silicon Valley CEOs, whom lawmakers have criticized for historically declining to endorse specific legislative proposals.
“Sam Altman is night and day compared to other CEOs,” Blumenthal, who began the hearing with an audio clip mimicking his voice that he said was generated by artificial intelligence trained on his floor speeches, told reporters. “Not just in the words and the rhetoric but in actual actions and his willingness to participate and commit to specific action.”
Altman has done a clever PR job here, frightening the horses sufficiently that competitors will have to struggle through regulation that OpenAI can probably handle easily, because it’s ahead of them. Get the lawmakers happy with you, and you’re halfway there.
unique link to this extract
Apple previews new accessibility features, including AI-generated voice clone • Six Colors
Apple organizes its accessibility features and settings by functional categories: Vision, Hearing, Physical and Motor. Now there’s Speech, too. New features under the Speech heading support those who are partially or fully nonverbal. Personal Voice is an intriguing feature that might seem familiar to anyone who has experienced AI-based text-to-speech that’s been trained on an actual human voice.
Those diagnosed with ALS [amyotrophic lateral sclerosis, aka motor neurone disease] are at great risk for losing their ability to speak, but often have advance warning. Using Personal Voice, an individual will be able to use an Apple Silicon-equipped Mac, iPhone or iPad to create a voice that resembles their own. If the ability to speak is lost, text the user generates on the device can then be converted to voice, for use in a variety of ways. It will work with augmented communication apps that are often used to make it easier for people with limited speech to be understood. And no, you can’t create a new Siri voice this way. All Personal Voice training is done on-device.
Live Speech can use an existing Siri voice to give people with speech disabilities a quick way to use voice to express common phrases or sentences. Type and save a statement, like a food order or a greeting, then tap the text to have it spoken aloud. It works inside Phone and FaceTime, or in-person, and users can save common phrases.
Apple using AI, but in a way that isn’t saying “hey, look at our AI, which works on-device!”
Introducing Ask Skift, the AI chatbot for your travel questions • Skift
Today, we are very excited to announce Ask Skift, the AI chatbot answers engine specializing in the travel industry, here to be of service on your daily work queries.
Go ahead, ask away any question, such as “How is Airbnb planning to leverage AI?” Or “Who is the new CEO of IHG?”. Or “Who are the owners of Ace Hotel?” Or, “Write me a short essay on the state of overtourism post-covid.”
We have “trained” Ask Skift on all the sum totality of Skift archives over the last 11 years, including daily stories, research reports, all of our clients’ trends reports, our specialized products – Airline Weekly, Daily Lodging Report, and Skift Meetings – and all the U.S. public travel companies’ financial SEC annual and quarterly reports. As soon as a new story or report is published, it goes straight into Ask Skift.
And this is just the start: we will continue to train it on other specialized travel industry content and data in coming weeks in order to improve the answers and expand the universe of queries it can tackle.
For now we have built this on top of OpenAI’s GPT-3.5 deep learning artificial intelligence algorithm (a logic-learning machine, or LLMin short). We would really like to use GPT-4, which gives exponentially better answers (we know, we have been testing internally on both versions) but for now it is cost-prohibitive. We expect prices to come down later this year and will upgrade to it.
Ali, who started and runs Skift, is a very smart person in the media space: I believe he has the rare distinction of never having been in charge of a media site that has closed. And this – training a chatbot on the specialist content of your site – is a really clever wrinkle in the media landscape. What if you tuned a chatbot on The Guardian’s deep, deep content, or the NY Times’s, or any sufficiently longstanding media site?
unique link to this extract
“The questions raised in the document between our developer team and our privacy legal counsel appear a bit out of context. The issue raised was a two-part technical question related to timing and whether or not it was even possible for us to be in possession of this kind of data,” Lawrence said.
Hmm, perhaps. You’ll recall that Telly was going to have a big screen, and “inescapable” ads that would scroll (we presume) along a smaller screen below that. To which reader/commenter starbird2005 observed “I wonder why someone wouldn’t just put some black paper over the second monitor. Then the ads play but you’ll never see them. Sort of reminds me of that CueCat scanner, which turned out to be a very handy barcode scanner when you hacked it.” Paper! The best hack.
unique link to this extract
Skylight: forecast Golden Hour and Sunset
Here’s a clever twist on a photography app, from the makers of the iOS app Halide:
Skylight uses atmospheric information to give you a forecast for evening light.
So if you’re trying to catch that perfect sunset for your holiday pic, this will tell you what sort of sunset you’re going to get, plus the “Golden Hour” – the period when the light is loveliest before sunset.
You’ll go and look at them after you’ve taken them, right? Print them? Frame them?
unique link to this extract
A photographer embraces the alien logic of AI • The New Yorker
Chris Wiley spoke to the (art) photographer Charlie Engman, who started playing around with Midjourney:
“And then, one time, I randomly did make something that I was, like, This is maybe cooler than anything I’ve ever made. How did that happen?”
The image in question was of a pair of ginger-haired boys perched on a couch with what looks like a miniature horse. One boy is affectionately nuzzling the other’s face. There is a distant resemblance to Dorothea Tanning’s surrealistic scenes of people and their animals, perhaps, but on closer inspection Engman’s images reveal layers of A.I. oddities. One boy’s legs appear to be merging into the sofa, and his hands have too many fingers (an easy A.I. tell). The lower halves of the second boy and their equine companion both seem in the process of being swallowed up, like loose change, by a black hole between the couch cushions.
Engman continued to make variations on the “couch with a horse” theme, each one stranger than the next. Like the Midjourney program itself, which responds to prompts with batches of four images at a time, Engman as an A.I. artist is dizzyingly prolific. “The amazing thing about A.I. is that I can make, like, three hundred pictures a day,” he told me, “And every single one of them can be an entirely different set of characters, and new location, and new material. I’m not constrained by physical reality at all.”
Physical reality, of course, is something that A.I. is completely unfamiliar with, a fact that Engman exploits to his benefit. “There was a while where I really loved how it iterated bodies in space,” Engman told me, “I was, like, How does it understand how people sit in chairs? How does it understand how people hug each other?” One series of images he made shows contorted, malformed human figures sitting in and often merging with various chairs, like a freaky update of a series of hilarious photographs by the artist Bruno Munari titled “Seeking Comfort in an Uncomfortable Chair.” Another features groups of businesswomen amorously engaged with a motley collection of semi-humanoid inflatables. My favorite series shows groups of middle-aged suburbanites standing in parking lots kissing. Each configuration looks like a hybrid of an erotic-contact improv troupe and a swarm of feeding lampreys.
Notable how uninterested he is in the copyright issue that has so many others freaking out. “We are all trained on, like, everything”, he says.
unique link to this extract
In Norway, the electric vehicle future has already arrived • The New York Times
About 110 miles south of Oslo, along a highway lined with pine and birch trees, a shiny fueling station offers a glimpse of a future where electric vehicles rule.
Chargers far outnumber gasoline pumps at the service area operated by Circle K, a retail chain that got its start in Texas. During summer weekends, when Oslo residents flee to country cottages, the line to recharge sometimes backs up down the off-ramp.
Marit Bergsland, who works at the store, has had to learn how to help frustrated customers connect to chargers in addition to her regular duties flipping burgers and ringing up purchases of salty licorice, a popular treat. “Sometimes we have to give them a coffee to calm down,” she said.
Last year, 80% of new-car sales in Norway were electric, putting the country at the vanguard of the shift to battery-powered mobility. It has also turned Norway into an observatory for figuring out what the electric vehicle revolution might mean for the environment, workers and life in general. The country will end the sales of internal combustion engine cars in 2025.
Norway’s experience suggests that electric vehicles bring benefits without the dire consequences predicted by some critics. There are problems, of course, including unreliable chargers and long waits during periods of high demand. Auto dealers and retailers have had to adapt. The switch has reordered the auto industry, making Tesla the best-selling brand and marginalising established carmakers like Renault and Fiat.
But the air in Oslo, Norway’s capital, is measurably cleaner. The city is also quieter as noisier gasoline and diesel vehicles are scrapped. Oslo’s greenhouse gas emissions have fallen 30% since 2009, yet there has not been mass unemployment among gas station workers and the electrical grid has not collapsed.
The big losers? Car dealers. Of note from this story: Norwegian kindergarten children take their daytime nap outside, “weather permitting”. And Norway is a big fossil fuel exporter – an irony the Green Party there acknowledges.
unique link to this extract
What would happen if the Succession fire played out in real life • Slate
Richard L. Hasen on the plotline in the latest episode of Succession, where a fire in a ballot counting station in Democratic-leaning Milwaukee, Wisconsin, is thought to tip the state Republican, thus handing (perhaps?) the election to a weird Trump-like figure:
Wisconsin’s election statutes do not appear to speak to what would happen with the massive destruction of ballots on Election Day. Many states interpret vague election statutes to favor enfranchisement of the voter, but Wisconsin gives less protection for absentee ballots, as the key state Supreme Court justice in the 2020 case of Trump v. Biden explained. If the justices on the state Supreme Court divided along party lines, as is often (but not always) the case, thanks to the recent election of Janet Protasiewicz, the court likely would side with the left-leaning candidate and offer some kind of remedy. Doing so would prevent voter disenfranchisement. If the same scenario were to take place in a potential tipping-point state that had a more conservative-leaning state Supreme Court, such as North Carolina, however, it could go another way.
To carry on the hypothetical based on the premise of a divided state court with a pro-democracy lean, like in Wisconsin: Perhaps the state court would require a partial revote in Milwaukee, as was suggested by Shiv in the Succession episode and by Claire Woodall-Vogg, executive director of the Milwaukee Election Commission, who consulted on the Succession episode. Woodall-Vogg explained that election officials would have records to know whose absentee ballots were destroyed in the fire.
But a revote may violate federal law, which requires that there be a uniform day on Election Day. (My former dean Erwin Chemerinsky unsuccessfully tried to get a revote in Palm Beach County, Florida, in 2000 after many voters were misled to vote for Pat Buchanan rather than Al Gore by the infamous butterfly ballot.)
I’m glad he mentioned Bush-Gore, because that was an obvious example where vote counting was undermined and the courts were relied on to rule. America isn’t very good at this democracy game, if we’re honest about it.
(Don’t worry about this being a spoiler for the episode, because it’s much more about the interpersonal relationships.)
unique link to this extract
Vice Media: from Murdoch money to bankruptcy in a decade • Press Gazette
Joseph Teasdale, head of tech on Enders Analysis’ media team, told Press Gazette the problem was “Vice never figured out a model at all”.
“Vice had a pitch – we know how to engage young people – but they never found a way to turn that pitch into a business,” Teasdale said. “They tried digital advertising, sponsored content, creative agency work, TV production, but continually missed revenue targets and never hit sustained profitability.”
Jim Bilton, managing director at Wessenden Marketing, drew attention to the role the tech platforms had in Vice’s financial downfall. “Despite some interesting and quite clever diversifications, the core business model is ad-driven, volume-driven and ultimately dependent on the big tech platforms to deliver audiences that Vice will never own themselves – the reverse of what the smarter legacy media companies are doing,” Bilton said. “The bottom line is that the better legacy media organisations are actually much more agile, smarter and multi-dimensional than the ‘one trick pony’ Vice. Trusted brands, audience-appropriate content, quality independent journalism, tight management and common-sense should/must win in the long-term!”
Teasdale added that Vice, in common with Buzzfeed, had believed that their online content businesses would scale in a manner similar to the software and platform successes of the last decade.
“You invest up front, and if you grow users enough, your revenues will eventually vastly outstrip your costs. But journalism is a lot more of a widgets business than people thought: if you want people to keep coming to your site, you need to keep making content, and so you need to keep spending money. A news business like Buzzfeed or Vice could never enjoy the kind of margins a platform business like Facebook can.”
|• Why do social networks drive us a little mad?
• Why does angry content seem to dominate what we see?
• How much of a role do algorithms play in affecting what we see and do online?
• What can we do about it?
• Did Facebook have any inkling of what was coming in Myanmar in 2016?
Read Social Warming, my latest book, and find answers – and more.
Errata, corrigenda and ai no corrida: none notified