Speed of AI development is outpacing risk assessment

Traditional methods of evaluating accuracy and safety are flawed.

Logo montage

Enlarge / Google, Anthropic, Cohere, and Mistral have each released AI models over the past two months as they seek to unseat OpenAI from the top of public rankings. (credit: FT)

The increasing power of the latest artificial intelligence systems is stretching traditional evaluation methods to breaking point, posing a challenge to businesses and public bodies over how best to work with the fast-evolving technology.

Flaws in the evaluation criteria commonly used to gauge performance, accuracy, and safety are being exposed as more models come to market, according to people who build, test, and invest in AI tools. The traditional tools are easy to manipulate and too narrow for the complexity of the latest models, they said.

The accelerating technology race sparked by the 2022 release of OpenAI’s chatbot ChatGPT and fed by tens of billions of dollars from venture capitalists and big tech companies, such as Microsoft, Google, and Amazon, has obliterated many older yardsticks for assessing AI’s progress.

Read 23 remaining paragraphs | Comments

Kobo adds color to its e-reader lineup for the first time, starting at $149

New Kobos are a lot cheaper than most no-name color e-readers you can buy.

Color e-readers have been a thing for a while, but until now, the biggest companies with the most extensive book ecosystems—Amazon, mainly, but also Barnes & Noble and Rakuten Kobo—have only sold traditional black-and-white models.

That changes on April 30th, when Kobo releases its first color e-readers: the $149.99 Kobo Clara Colour and $219.99 Kobo Libra Colour. Both devices look a lot like their black-and-white predecessors, the Kobo Libra 2 and Kobo Clara 2E, but with colorful screens instead of black-and-white ones.

Kobo is also refreshing the black-and-white version of the Clara, called the Clara BW to distinguish it from the color model. It's mostly identical to the old Clara 2E model, but with the faster dual-core processor from the color models. It sells for $129.99, $10 cheaper than the Clara 2E.

Read 5 remaining paragraphs | Comments

Joaquin Phoenix meets his perfect match in Joker: Folie à Deux teaser

“I’m not alone anymore”: Lady Gaga costars as Harley Quinn.

Joaquin Phoenix returns as Arthur Fleck in Joker: Folie à Deux.

Joaquin Phoenix won an Oscar for his portrayal of a failed stand-up comedian struggling with mental illness in the 2019 film Joker, director Todd Phillips' controversial interpretation of the classic Batman villain. The honor was richly deserved. In my review, I called it a "masterful performance" that "transforms the narrative into something more than a competent-but-unremarkable tale of hard knocks driving a troubled man to violence." The film ended up topping $1 billion globally at the box office—the first R-rated movie to do so, making it the highest-grossing R-rated film ever.

Now Phoenix is reprising that role in Phillips' follow-up, Joker: Folie à Deux, costarring Lady Gaga as Harley Quinn. It's hard to imagine Phillips matching his earlier achievement, but Warner Bros. released the first teaser last night, and quite frankly, the film looks fantastic.

(Spoilers for the 2019 film below.)

Read 6 remaining paragraphs | Comments

Mitbestimmung: Gericht entscheidet über ChatGPT-Nutzung im Betrieb

Das Arbeitsgericht Hamburg hat entschieden, dass Betriebsräte kein Mitbestimmungsrecht haben, wenn der Arbeitgeber die Nutzung von ChatGPT über private Accounts der Mitarbeiter zulässt. (ChatGPT, KI)

Das Arbeitsgericht Hamburg hat entschieden, dass Betriebsräte kein Mitbestimmungsrecht haben, wenn der Arbeitgeber die Nutzung von ChatGPT über private Accounts der Mitarbeiter zulässt. (ChatGPT, KI)

The most metal of rockets has gone into the great mosh pit in the sky

The Delta IV booster seems hardly like a champion for commercial launch, but here we are.

I've got a guilty secret that I can now share—I loved the Delta IV Heavy rocket.

No, I didn't love the price, which was preposterous, at times approaching $400 million. This precluded Delta from having any other customers than the US government. I didn't love the low flight rate, just 16 missions in 20 years. This prevented the rocket's operator, United Launch Alliance, from ever approaching anything remotely like efficient operations.

But there were two things I adored about the Delta IV Heavy rocket, which made its final launch on Tuesday. I loved watching it take flight. And I love that, warts and all, it demonstrated that private companies could develop a heavy lift rocket. The Delta booster, although the product of decades of traditional space development, offered a glimpse of the commercial launch future that we're living in today.

Read 14 remaining paragraphs | Comments

Start der Fallout-Serie: Wenn der Atompilz so groß wie dein Daumen ist, renn!

Die Erwartungen an die Verfilmung der Games waren groß. Nun läuft die Fallout-Serie endlich an, startet gleich mit einer extrem starken Szene – und liefert auch sonst ab. Eine Rezension von Peter Osteried (Science-Fiction, Amazon)

Die Erwartungen an die Verfilmung der Games waren groß. Nun läuft die Fallout-Serie endlich an, startet gleich mit einer extrem starken Szene - und liefert auch sonst ab. Eine Rezension von Peter Osteried (Science-Fiction, Amazon)