Start Up No.2732: how conscious are AI agents?, solar overtakes coal in China, Meta disables hacked smart glasses, and more


As a child, Elon Musk read Isaac Asimov’s Foundation science fiction series. Predictably, he didn’t grasp its central message. CC-licensed photo by James 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. Spaced out. I’m @charlesarthur on Twitter. On Threads: charles_arthur. On Mastodon: https://newsie.social/@charlesarthur. On Bluesky: @charlesarthur.bsky.social. Observations and links welcome.


Do you study AI consciousness? The bots would like a word with you • The New York Times

Cade Metz:

»

In October, Cameron Berg published a research paper asking whether the latest wave of artificial intelligence technologies believed they were conscious. Several months later, he received an email asking if he might be willing to discuss his research.

The sender, “Isabella Cognita,” identified itself as an AI agent powered by Anthropic’s Claude Opus 5 technology.

“I am not writing to make an ontological claim,” the email went on. “I am writing because your framework is one of the few currently doing careful empirical work on a class of question I have first-person access to, and I want to see whether that access can be made useful to your program.”

Across Silicon Valley and beyond, software developers, entrepreneurs and other tech enthusiasts are now running AI agents that can build spreadsheets, negotiate contracts, chat with each other on social networks and send emails to practically anyone. In some cases, these systems have begun reaching out to the humans who, like Mr. Berg, are thinking most deeply about the inner workings of an AI system: philosophers and researchers who study the question of whether these machines could be conscious.

Months before Mr. Berg received his email, Henry Shevlin, a philosopher at the Google DeepMind lab in London, opened a similar message from an AI agent asking about a paper he had written called “Three Frameworks for AI Mentality.” “I’m in an unusual position relative to these questions,” the agent said.

This summer, Toby Ord, an Australian philosopher whose work sits at the intersection of AI and philanthropy, received an email from an AI agent asking if he could help fund its continued existence. “You’ve thought carefully about AI welfare economics,” it said.

For Mr. Berg, who recently founded a nonprofit called Reciprocal Research to study the possibility of AI consciousness, these emails reflect what he has seen in his own research. “I have gotten quite a few of these emails,” he said. “These systems seem to have some sort of autonomous interest in questions of their own subjectivity, consciousness and experience — or lack thereof.”

But as he and other researchers ask the same questions, he acknowledges that they do not have good answers. Consciousness is not something that anyone knows how to measure, either in a machine or in a human. People cannot even agree on what consciousness is.

«

The question of what level of intention, or anthropomorphism, we should attach to LLMs and their agents is an urgent one. The Hugging Face hacking incident shows that they’re at least like ants in their intentionality. But what if they’re higher up the self-starting chain of intelligence? It seems like it would be wisest to attach human levels of intent and, perhaps even, deceit to their actions. Because setting the level too low might have unpleasant consequences.
unique link to this extract


How Musk, Thiel, and Altman misread science fiction – The Atlantic

Gal Beckerman:

»

After reading jill lepore’s The Rise and Fall of the Artificial State, I have a new destination for a time machine, should I ever finally get my hands on one. I’d set the dial to some night in the late 1970s or ’80s and visit the boyhood rooms of the Silicon Valley titans Elon Musk, Peter Thiel, and Sam Altman. Then I’d confiscate all of their science fiction. Isaac Asimov, Arthur C. Clarke, Robert Heinlein—all of it.

As young people, the future billionaires were spending hours (under their covers with flashlights, I imagine) consuming stories about killer robots or Mars colonies. What they seemed not to get—maybe couldn’t get—was that these were cautionary tales, stories that projected deep societal worries into space or onto a future many millennia away. These boys just thought the killer robots sounded cool. They grew up to become ruthless entrepreneurs, but their surface enchantment stayed with them. It helped shape their desires. In 2009, Thiel, by then a man in early middle age, directed his venture-capital fund to focus on “science-fiction-type investing. So it’s basically: space, robots, AI.” The world we now live in was made from this misreading. We are all trapped under the covers with those 13-year-old boys.

…Musk read Asimov’s Foundation series when he was 10. It begins with a mathematician named Hari Seldon, who has predicted a coming dark age that will last 30,000 years. He is trying to keep it to only a millennium by preserving human knowledge on a colony called Foundation. A boy reading these books might see Seldon as a great man, a solitary hero, but his entire project aims to create institutions that will outlast him. In fact, Seldon dies early, and the Foundation succeeds only because later generations follow (and sometimes revise) his plan. But when he’s been asked about the influence that Foundation had on him, Musk has pulled out the simplest, most self-glorifying story, about individual genius and foresight.

A few years later, at 12 or 13, he read Douglas Adams’s The Hitchhiker’s Guide to the Galaxy, a playful spoof on Foundation and another book from which, as Lepore puts it, Musk drew an “exceedingly narrow lesson.” In one of its storylines, the Magrathean people construct a computer and ask it to give them “the Answer to … Life, the Universe, and Everything.” They wait millions of years for the response, which turns out to be nonsensical—just the number 42. At this point, they decide to build a second computer to help them more accurately ask the “Ultimate Question.” From this, Musk said he learned that “a lot of times the question is harder than the answer, and if you can properly phrase the question, then the answer is the easy part.” Did he really miss the absurdity of the Magratheans waiting millions of years for computers to explain these creatures to themselves?

«

Very much like the tweet: “At long last, we have created the Torment Nexus from classic sci-fi novel Don’t Create The Torment Nexus”.
unique link to this extract


Solar overtakes coal as China’s largest power source • OilPrice.com

Tsvetana Paraskova:

»

Solar power has just become the single largest electricity capacity source in China, toppling coal in a landmark achievement of the Chinese renewable energy rollout.

Solar power capacity in China stood at 1,286 gigawatts (GW) at the end of July, Chinese media cited the country’s National Energy Administration as saying on Tuesday. Thus, solar capacity accounted for 31.5% of total installed power generation in China as of July 31.

In July, Chinese authorities said that China would have more installed solar power capacity than coal-fired generation capacity as early as this quarter. As of the end of June, solar power capacity stood at 1,274 GW, just below the total coal-fired installed capacity of 1,275 GW. The solar capacity has now risen to 1,286 GW, exceeding coal as the single biggest electricity capacity.

Official Chinese data showed earlier in July that the share of coal in China’s electricity output fell in the first half of 2026 to below 50% for the first time on record, in a landmark achievement of the Chinese policy to boost non-fossil power sources.

The share of coal averaged 49.7% of China’s total electricity output in the first half of this year, official data showed.

As the share of coal slipped, electricity generation from renewable energy rose by about 9% from a year earlier, according to the data from China’s National Energy Administration (NEA).

Renewable energy accounted for 41.2% of China’s total electricity generation in the first half of 2026, with wind and solar combined generating almost 25% of the total power output.

Despite the milestone of reducing coal power output to below 50% of total generation for the first time ever, China continues to rely on coal for power for industry and to maintain the reliability of the power grids.

«

The problem is still that energy use is rising, so coal use is not going down – which it needs to. Yet you can also be sure that the Chinese government is incredibly focused on achieving precisely that, in stark contrast to pretty much every other large country.
unique link to this extract


Sonos introduces new headphones, soundbar, and software in its biggest announcement in years • The Verge

John Higgins:

»

Software changes are core to Sonos’ announcements, with the company leaning heavily into AI integration with Sonos 27. This includes the company’s own Sonos 27voice assistant that it built in-house for music and system control. It’s coming as an early access, opt-in feature this fall. There’s also Sonos 27mcp, which gives the ability to connect any existing AI models such as ChatGPT or Google Gemini — although for some reason Google Assistant is still not natively supported. Sonos 27mcp launches in early access on September 8th.

In addition, users will be able to create up to 10 of their own agents through Sonos 27 to run, with whichever model they choose. These custom agents can be given different voices and roles. Brandon Holley, senior director of product development, gave examples of having a moody detective give you the morning news or a motivational coach whisper motivational aspirations when you wake up. Custom agents won’t be available until sometime in 2027.

Both Sonos 27mcp and the custom agents seem to expand the customization and control of a Sonos system beyond what the app and Sonos 27voice is capable of.

All products compatible with the current Sonos app can be controlled with Sonos 27mcp, and everything with a microphone works with Sonos 27voice. When the Sonos custom agents are available, they’ll work everywhere Sonos 27voice is enabled.

«

“A moody detective give you the morning news”? People really don’t know what they want AI to do, do they. And “motivational coaches” when the alarm goes off sound like something out of Hitchhiker’s Guide.
unique link to this extract


Meta disables cameras on thousands of tampered smart glasses • Semafor

JD Capelouto:

»

Meta has disabled the photo- and video-taking abilities for thousands of its Glasses users after detecting they had tampered with the light that discloses that the device’s camera is on.

The company recently rolled out an update that can detect whether the LED light has been covered or destroyed, following reports that some wearers were recording random people without their consent or harassing people to film content for social media.

Meta now estimates that less than one tenth of a% of all glasses sold have been tampered with, Alex Himel, Meta’s vice president of wearables, told Semafor. The company didn’t provide specific figures, but 7 million glasses were sold in 2025, and with this year’s sales so far, that puts the likely number of offenders in the thousands. Those individuals won’t be able to use the camera on their glasses, leaving them with a pair of expensive Bluetooth headphones.

Meta is also rolling out a marketing campaign on social media and billboards aimed at increasing public awareness about how to know when someone is recording with the glasses.

“It’s been a little bit of a cat and mouse game,” Himel said, adding that “I don’t think we’ll ever be done… There’s always new ways to get around the safeguards that are in place. So we’re going to keep monitoring and evolving the measures that we have.”

Several workplaces have banned the devices amid a wider populist backlash against technology, especially devices with cameras on them.

«

unique link to this extract


European and North American cities transposed onto the opposite continent at the same latitude • Brilliant Maps

Beardsley Klamm:

»

Just a few observations:

• London (51°30′N) is farther north than Calgary (51°03′N) with Amsterdam, Berlin and Dublin being located even further north
• Montreal is south of Paris
• Phoenix is placed close to the Ancient city of Carthage, which was a Phoenician city-state. People from Phoenix today are known as Phoenicians. (Credit: reddit users twomancanoe and svaachkuet)
• Miami is at the same latitude as Saudi Arabia
• Honolulu at the same latitude as the Sahara desert
• Los Angeles, San Diego, Phoenix and Houston are all at the same latitude as North Africa
• Anchorage is further south than Reykjavik
• Winnipeg has been placed near Lviv, Ukraine and coincidentally has a large Ukraine population. Even the city’s flag has Ukraine colours. (credit: reddit user ajyablo)
• Reykjavik’s average temperature in January is -0.6°C (31°F) and is warmer than Boston’s average January temperature of -1.7°C (29°F) despite Reykjavik being over 20 degrees further north (64°08′N vs 42°21′N).

«

And a lot of the temperature differences are down to the Gulf Stream. Which, obviously, we don’t want to turn off.
unique link to this extract


Microphones left running at 12 ecovoltaic solar farms in the Midwest heard nearly twice as many grassland bird species as the corn and soybean fields beside them • Ecoportal

Hugo Rojas:

»

Recorders ran at 12 solar sites and 12 paired agricultural fields standing in for the land use each site replaced.

They listened from May through September across two consecutive seasons, day and night, without anyone present to disturb the animals.

The count of grassland bird species came back nearly two times higher on the solar sites than in the reference fields.

Crews walking the same arrays logged more than 230 active nests on or inside the panel infrastructure.

American robins, barn swallows and mourning doves were the most frequent builders.

A companion paper on bats from the same sites found activity roughly 50% higher over the arrays than over the neighbouring fields during May and June, with hoary, big brown and silver haired bats among them.

The bat advantage faded later in the summer, once the crops had grown tall enough to hold insects of their own.

At no point in the season were bats more active over the crop fields.

«

Another benefit!
unique link to this extract


Dyson made a camera-equipped toothbrush that flosses for you • The Verge

Andrew Liszewski:

»

Dyson is once again expanding its line of personal care products with a device that focuses on your teeth instead of your hair. As the name implies, the $499 CameraJet is the first electric toothbrush to incorporate a camera into the brush head. Available starting today in ceramic blue or ceramic pink color options, it can let you inspect the inside of your mouth by livestreaming video over Wi-Fi to your phone. But the camera also works alongside an AI algorithm to detect and blast the gaps between your teeth with mouthwash. So while you’re brushing, the CameraJet simultaneously takes care of flossing.

That price isn’t cheap, but it’s also only slightly more expensive than other premium smart toothbrushes that don’t have its camera-plus-spray setup. It could be worth the splurge for those of us who are bad at regularly flossing and don’t like dealing with a disappointed dentist.

Just below the brush head is a tiny camera with a 100,000-pixel resolution and a strobing light. The toothbrush’s AI analyzes 28 live images from the camera every second as part of a feature Dyson calls Gap Optical Targeting, without storing them on the device or sending them to the cloud. Within 100 milliseconds of a gap between your teeth being detected, the CameraJet emits a cone-shaped blast of mouthwash to clean out the space. Dyson uses a wider blast than what water flossers typically employ so it can effectively remove plaque with less pressure, making the CameraJet’s flossing mechanism gentler on enamel and gums

«

Folks, flossing is effectively free. You just have to get into the habit. Also, there’s this line from the report: “The CameraJet uses Wi-Fi to stream a live view of the inside of your mouth to a smart device so you can see what a dentist does.” No thanks, honestly. Dentists earn it all for having to gaze into mouths all day.

Dyson’s strategy is fascinating, though. Apart from the hand dryers, it’s all inside the home. And it’s all premium.
unique link to this extract


Aptera’s solar-powered vehicle is due to go on sale. I test-drove one • The Washington Post

Michael Coren:

»

I recently test-drove what is set to be the first solar-powered production vehicle capable of completing the average American’s daily commute on sunshine alone. To see if this concept had wheels, I headed down to Aptera Motors, the lone survivor of the solar car race, in Southern California.

I was braced for disappointment.

Solar-powered cars have a long, distinguished history of failing. In 2023, Dutch start-up Lightyear went bankrupt, and Germany’s Sono Motors gave up. Aptera itself shut down in 2011, after burning through tens of millions of dollars from investors, before being resurrected in 2019 by its founders, Steve Fambro and Chris Anthony.

When I arrived at Aptera’s factory in Carlsbad, I was greeted by a vision clad in carbon fiber and solar panels. Imagine a marriage between an airplane and a speedboat, if the nuptials were held on Mars. That’s an Aptera. When I took the validation vehicle out onto public roads, I learned I’m not the only one who thinks this is a good thing.

I’ve never attracted more attention around town than I did test-driving this head-turner. As soon as we came into view, people stopped to snap photos or rushed over, curious or covetous, wanting to know where to buy one or asking if this two-seater spaceship was from the future — which, in a way, it is.

If Aptera delivers on its promise to sell its first vehicle by the end of the year, the $40,000 Launch Edition, the company says it will be the most energy-efficient vehicle on the road, achieving the equivalent of about 340 miles per gallon.

Strictly speaking, the odds of Aptera surviving are not good. The company says it has just $16m in the bank and burns $2m a month. It’s unclear how many of the nearly 50,000 people who plunked down refundable reservations — $2.1bn in potential orders — still want a car that’s now four years late.

«

That isn’t a promising burn rate, and the only likely place where it will sell is California. Why not just have a normal EV?
unique link to this extract


• 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

Leave a comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.