Month: December 2023

Enlarge

Google announced big changes to its most legally fraught set of user settings: your location data. Google's misleading Location History descriptions in Google Maps have earned it several lawsuits in the US and worldwide. A quick count involves individual lawsuits in California, Arizona, Washington, a joint lawsuit in Texas, Indiana, and the District of Columbia, and another joint lawsuit across 40 additional US states. Internationally, Google has also been sued in Australia over its location settings. The point is that any change to Google's location settings must have some motive behind it, so bear with us while we try to decode everything.

Google's big new location data change is a new, duplicate data store that will live exclusively on your device. Google's new blog post says data for the long-running Google Maps Timeline feature will now "be saved right on your device—giving you even more control over your data." That's right, one of the world's biggest Internet data companies advocates for local storage of your location data.

The company continues, "If you’re getting a new phone or are worried about losing your existing one, you can always choose to back up your data to the cloud so it doesn’t get lost. We’ll automatically encrypt your backed-up data so no one can read it, including Google." Users will apparently have lots of control over this new locally stored data, with Google saying, "Soon, you’ll be able to see all your recent activity on Maps... in one central place, and easily delete your searches, directions, visits, and shares with just a few taps. The ability to delete place-related activity from Maps starts rolling out on Android and iOS in the coming weeks."

Enlarge (credit: Getty Images)

On Wednesday, news quickly spread on social media about a new enabled-by-default Dropbox setting that shares your Dropbox data with OpenAI for an experimental AI-powered search feature. Dropbox says that user data shared with third-party AI partners isn't used to train AI models and is deleted within 30 days.

Even with assurances of data privacy laid out by Dropbox on an AI privacy FAQ page, the discovery that the setting had been enabled by default upset some Dropbox users. The setting was first noticed by writer Winifred Burton, who shared information about the Third-party AI setting through Bluesky on Tuesday, and frequent AI critic Karla Ortiz shared more information about it on X.

Ortiz expressed worries that the data might be trained secretly without consent. In its FAQ, Dropbox contradicts this claim, saying, "We won’t let our third-party partners train their models on our user data without consent."

Enlarge (credit: Getty Images)

On Wednesday, news quickly spread on social media about a new enabled-by-default Dropbox setting that shares your Dropbox data with OpenAI for an experimental AI-powered search feature. Dropbox says that user data shared with third-party AI partners isn't used to train AI models and is deleted within 30 days.

Even with assurances of data privacy laid out by Dropbox on an AI privacy FAQ page, the discovery that the setting had been enabled by default upset some Dropbox users. The setting was first noticed by writer Winifred Burton, who shared information about the Third-party AI setting through Bluesky on Tuesday, and frequent AI critic Karla Ortiz shared more information about it on X.

Ortiz expressed worries that the data might be trained secretly without consent. In its FAQ, Dropbox contradicts this claim, saying, "We won’t let our third-party partners train their models on our user data without consent."

Enlarge / Sultan Ahmed Al Jaber defied expectations to deliver a document that explicitly calls for limits on fossil fuel use. (credit: Fadel Dawod)

On Wednesday, the UN's COP28 meeting wrapped up with a major success: Despite a bruising fight with OPEC nations, the closing agreement included a call for a transition away from fossil fuels. There's still plenty here for various parties to dislike, but this is the first agreement that makes the implications of the Paris Treaty explicit: We can't limit climate change and continue to burn fossil fuels at anything close to the rate we currently do.

Beyond that, however, the report has something to disappoint everyone. It catalogs strong signs of incremental progress toward the Paris goals while acknowledging we're running out of time for further increments. And the steps it calls for will likely keep changes on a similar trajectory.

Taking stock

The new document is called a "Global Stocktake" in reference to checking the world's progress toward the goals of the Paris Agreement: limit climate change to 2° C above preindustrial temperatures and try to keep it to 1.5° C. That agreement called for nations to make pledges to limit greenhouse gas emissions; initial pledges were insufficient, but regular meetings of the Conference of the Parties (COP) would allow the pledges to be updated, raising their aggressiveness until the world is on a trajectory toward meeting its goals.

Enlarge / Sultan Ahmed Al Jaber defied expectations to deliver a document that explicitly calls for limits on fossil fuel use. (credit: Fadel Dawod)

On Wednesday, the UN's COP28 meeting wrapped up with a major success: Despite a bruising fight with OPEC nations, the closing agreement included a call for a transition away from fossil fuels. There's still plenty here for various parties to dislike, but this is the first agreement that makes the implications of the Paris Treaty explicit: We can't limit climate change and continue to burn fossil fuels at anything close to the rate we currently do.

Beyond that, however, the report has something to disappoint everyone. It catalogs strong signs of incremental progress toward the Paris goals while acknowledging we're running out of time for further increments. And the steps it calls for will likely keep changes on a similar trajectory.

Taking stock

The new document is called a "Global Stocktake" in reference to checking the world's progress toward the goals of the Paris Agreement: limit climate change to 2° C above preindustrial temperatures and try to keep it to 1.5° C. That agreement called for nations to make pledges to limit greenhouse gas emissions; initial pledges were insufficient, but regular meetings of the Conference of the Parties (COP) would allow the pledges to be updated, raising their aggressiveness until the world is on a trajectory toward meeting its goals.

Enlarge (credit: Leon Neal / Staff | Getty Images Europe)

It's hard to know exactly how dire the financial situation is at Elon Musk's X (formerly Twitter). However, insider sources recently revealed to Bloomberg that the social media platform expects to end 2023 with "roughly" $2.5 billion in advertising revenue.

That's "a significant slump from prior years," sources said. It's also about half a billion short of the $3 billion that X executives expected to make in ad sales in 2023, one source said.

Last year, Twitter raked in more than $1 billion in ad revenue per quarter, sources said. But in each of the first three quarters of 2023, X only managed to generate "a little more than $600 million" in ad revenue. Now, the most recent advertiser fallout over antisemitic content on X—estimated in November as triggering a sudden $75 million loss—is still casting a shadow on what could become an even more dismal fourth quarter.

Enlarge (credit: Leon Neal / Staff | Getty Images Europe)

It's hard to know exactly how dire the financial situation is at Elon Musk's X (formerly Twitter). However, insider sources recently revealed to Bloomberg that the social media platform expects to end 2023 with "roughly" $2.5 billion in advertising revenue.

That's "a significant slump from prior years," sources said. It's also about half a billion short of the $3 billion that X executives expected to make in ad sales in 2023, one source said.

Last year, Twitter raked in more than $1 billion in ad revenue per quarter, sources said. But in each of the first three quarters of 2023, X only managed to generate "a little more than $600 million" in ad revenue. Now, the most recent advertiser fallout over antisemitic content on X—estimated in November as triggering a sudden $75 million loss—is still casting a shadow on what could become an even more dismal fourth quarter.

Enlarge / A Rock'em Sock'em AI model battle. (credit: CSA Images)

As the AI landscape has expanded to include dozens of distinct large language models (LLMs), debates over which model provides the "best" answers for any given prompt have also proliferated (Ars has even delved into these kinds of debates a few times in recent months). For those looking for a more rigorous way of comparing various models, the folks over at the Large Model Systems Organization (LMSys) have set up Chatbot Arena, a platform for generating Elo-style rankings for LLMs based on a crowdsourced blind-testing website.

Chatbot Arena users can enter any prompt they can think of into the site's form to see side-by-side responses from two randomly selected models. The identity of each model is initially hidden, and results are voided if the model reveals its identity in the response itself.

The user then gets to pick which model provided what they judge to be the "better" result, with additional options for a "tie" or "both are bad." Only after providing a pairwise ranking does the user get to see which models they were judging, though a separate "side-by-side" section of the site lets users pick two specific models to compare (without the ability to contribute a vote on the result).

Enlarge / A Rock'em Sock'em AI model battle. (credit: CSA Images)

As the AI landscape has expanded to include dozens of distinct large language models (LLMs), debates over which model provides the "best" answers for any given prompt have also proliferated (Ars has even delved into these kinds of debates a few times in recent months). For those looking for a more rigorous way of comparing various models, the folks over at the Large Model Systems Organization (LMSys) have set up Chatbot Arena, a platform for generating Elo-style rankings for LLMs based on a crowdsourced blind-testing website.

Chatbot Arena users can enter any prompt they can think of into the site's form to see side-by-side responses from two randomly selected models. The identity of each model is initially hidden, and results are voided if the model reveals its identity in the response itself.

The user then gets to pick which model provided what they judge to be the "better" result, with additional options for a "tie" or "both are bad." Only after providing a pairwise ranking does the user get to see which models they were judging, though a separate "side-by-side" section of the site lets users pick two specific models to compare (without the ability to contribute a vote on the result).