“Nauseatingly frightening”: Law firm condemns careless AI use in court. //

"As all lawyers know (or should know), it has been documented that AI sometimes invents case law, complete with fabricated citations, holdings, and even direct quotes," his letter said. "As we previously instructed you, if you use AI to identify cases for citation, every case must be independently verified."

A computer can never be held accountable. This legendary page from an internal IBM training in 1979 could not be more appropriate for our new age of AI.

A COMPUTER CAN NEVER BE HELD ACCOUNTABLE. THEREFORE A COMPUTER MUST NEVER MAKE A MANAGEMENT DECISION

A computer can never be held accountable

Therefore a computer must never make a management decision

When asked about topics such as the Tiananmen Square massacre, persecution of Uyghur Muslims, or Taiwan’s sovereignty, DeepSeek either dodges the question or parrots Beijing’s official rhetoric. This is not a bug—it’s a feature. Unlike Western AI models, which, for all their flaws, still allow for a broader range of discourse, DeepSeek operates within strict ideological parameters. It’s a stark reminder that AI is only as objective as the people—or governments—who control it. //

The question we must ask ourselves is simple: If AI can be programmed to push a state-sponsored narrative in China, what’s stopping corporations, activist organizations, or even Western governments from doing the same?

Don’t think American companies would stop at weighting their algorithms to ensure diversity. Over the past few years, we’ve seen a growing trend of corporations aligning themselves with Environmental, Social, and Governance (ESG) metrics. This framework prioritizes social justice causes and other politically charged issues, distorting how companies operate. Over the same period of time, many social media companies have taken aggressive steps to suppress content considered “misinformation.”. //

Without transparency and accountability, AI could become the most powerful propaganda tool in human history—capable of filtering search results, rewriting history, and nudging societies toward preordained conclusions. //

This moment demands vigilance. The public must recognize the power AI has over the flow of information and remain skeptical of models that show signs of ideological manipulation. Scrutiny should not be reserved only for AI developed in adversarial nations but also for models created by major tech companies in the United States and Europe. //

DeepSeek has provided a glimpse into a world where AI is used to enforce state-approved narratives. If we fail to confront this issue now, we may wake up in a future where AI doesn’t just provide answers—it decides which questions are even allowed to be asked.

Over the millennia, we have created security systems to deal with the sorts of mistakes humans commonly make. //

But it’s not the frequency or severity of AI systems’ mistakes that differentiates them from human mistakes. It’s their weirdness. AI systems do not make mistakes in the same ways that humans do.

Much of the friction—and risk—associated with our use of AI arise from that difference. We need to invent new security systems that adapt to these differences and prevent harm from AI mistakes. //

AI errors come at seemingly random times, without any clustering around particular topics. LLM mistakes tend to be more evenly distributed through the knowledge space. A model might be equally likely to make a mistake on a calculus question as it is to propose that cabbages eat goats.

And AI mistakes aren’t accompanied by ignorance. A LLM will be just as confident when saying something completely wrong—and obviously so, to a human—as it will be when saying something true. The seemingly random inconsistency of LLMs makes it hard to trust their reasoning in complex, multi-step problems. If you want to use an AI model to help with a business problem, it’s not enough to see that it understands what factors make a product profitable; you need to be sure it won’t forget what money is. //

Matt • January 21, 2025 11:54 AM

“Technologies like large language models (LLMs) can perform many cognitive tasks”

No, they can’t perform ANY cognitive tasks. They do not cogitate. They do not think and are not capable of reasoning. They are nothing more than word-prediction engines. (This is not the same as saying they are useless.)

You should know better than that, Bruce.

RealFakeNews • January 21, 2025 12:35 PM

Part of the problem is AI can’t fundamentally differentiate a fact from something it just made up. It can check cabbages and goats are related via some probability, but it can’t check that a cabbage doesn’t eat goats because it can’t use the lack of data to verify if that is correct.

Changing just 0.001% of inputs to misinformation makes the AI less accurate.

A/I generated faces

Upload your photo and get a thorough, three-paragraph description of it. //

wanted to develop an alternative service for storing and sharing photos that is open source and end-to-end encrypted. Something “more private, wholesome, and trustworthy,” he says. The paid service he designed, Ente, is profitable and says it has more than 100,000 users, many of whom are already part of the privacy-obsessed crowd. But Mohandas struggled to articulate to wider audiences why they should reconsider relying on Google Photos, despite all the conveniences it offers.

Then one weekend in May, an intern at Ente came up with an idea: Give people a sense of what some of Google’s AI models can learn from studying images. Last month, Ente launched https://Theyseeyourphotos.com, a website and marketing stunt designed to turn Google’s technology against itself. People can upload any photo to the website, which is then sent to a Google Cloud computer vision program that writes a startlingly thorough three-paragraph description of it. (Ente prompts the AI model to document small details in the uploaded images.)

Hacker Uno Ars Centurion
7y
314
Subscriptor++
42Kodiak42 said:
Remember, a big enough privacy violation also constitutes a grave security vulnerability.
Technically, any privacy violation constitutes a grave security vulnerability.

Remember, confidentiality is one of the five fundamental security tenants, and it defends against unauthorized disclosure. When you violate privacy, you are committing an unauthorized disclosure.

For the record, the five fundamental security tenants are:

• Confidentiality, which defends against unauthorized disclosure of a protected asset.
• Integrity, which defends against unauthorized modification of a protected asset.
• Availability, which defends against denial of authorized access to a protected asset.
• Authenticity, which defends against spoofing, forgery, and repudiation of a protected asset.
• Access-Control, which defends against unauthorized access of a protected asset.

FrontierMath's difficult questions remain unpublished so that AI companies can't train against it. //

On Friday, research organization Epoch AI released FrontierMath, a new mathematics benchmark that has been turning heads in the AI world because it contains hundreds of expert-level problems that leading AI models solve less than 2 percent of the time, according to Epoch AI. The benchmark tests AI language models (such as GPT-4o, which powers ChatGPT) against original mathematics problems that typically require hours or days for specialist mathematicians to complete.

FrontierMath's performance results, revealed in a preprint research paper, paint a stark picture of current AI model limitations. Even with access to Python environments for testing and verification, top models like Claude 3.5 Sonnet, GPT-4o, o1-preview, and Gemini 1.5 Pro scored extremely poorly. This contrasts with their high performance on simpler math benchmarks—many models now score above 90 percent on tests like GSM8K and MATH.

The design of FrontierMath differs from many existing AI benchmarks because the problem set remains private and unpublished to prevent data contamination. Many existing AI models are trained on other test problem datasets, allowing the AI models to easily solve the problems and appear more generally capable than they actually are. Many experts cite this as evidence that current large language models (LLMs) are poor generalist learners.

Goldman noted that Ranson relying on Copilot for "what was essentially a numerical computation was especially puzzling because of generative AI's known hallucinatory tendencies, which makes numerical computations untrustworthy." //

Because Ranson was so bad at explaining how Copilot works, Schopf took the extra time to actually try to use Copilot to generate the estimates that Ranson got—and he could not.

Each time, the court entered the same query into Copilot—"Can you calculate the value of $250,000 invested in the Vanguard Balanced Index Fund from December 31, 2004 through January 31, 2021?"—and each time Copilot generated a slightly different answer.

This "calls into question the reliability and accuracy of Copilot to generate evidence to be relied upon in a court proceeding," Schopf wrote. //

Until a bright-line rule exists telling courts when to accept AI-generated testimony, Schopf suggested that courts should require disclosures from lawyers to stop chatbot-spouted inadmissible testimony from disrupting the legal system. //

Goldman suggested that Ranson did not seemingly spare much effort by employing Copilot in a way that seemed to damage his credibility in court.

"It would not have been difficult for the expert to pull the necessary data directly from primary sources, so the process didn't even save much time—but that shortcut came at the cost of the expert's credibility," Goldman told Ars.

American Deplorable ™
10 hours ago
A deep fake outlawing deep fakes.

The irony is almost as thick as the hair gel.

Because apps talking like pirates and creating ASCII art never gets old

LY Corp's QA team struggled to manage projects while wading through prolix posts

Create the most realistic speech with our AI audio platform

Pioneering research in Text to Speech, AI Voice Generator, and more

Generate high quality speech in any voice, style, and language. Our AI voice generator renders human intonation and inflections with exceptional fidelity, adjusting the delivery based on context.

Large language models are increasingly integrating into everyday life—as chatbots, digital assistants, and internet search guides, for example. These artificial intelligence systems, which consume large amounts of text data to learn associations, can create all sorts of written material when prompted and can ably converse with users.

Large language models’ growing power and omnipresence mean that they exert increasing influence on society and culture.

So, it’s of great import that these artificial intelligence systems remain neutral when it comes to complicated political issues. Unfortunately, according to a new analysis recently published to PLOS ONE, this doesn’t seem to be the case.

As Phil Root, the deputy director of the Defense Sciences Office at DARPA, recounted to Scharre, “A tank looks like a tank, even when it’s moving. A human when walking looks different than a human standing. A human with a weapon looks different.”

In order to train the artificial intelligence, it needed data in the form of a squad of Marines spending six days walking around in front of it. On the seventh day, though, it was time to put the machine to the test.

“If any Marines could get all the way in and touch this robot without being detected, they would win. I wanted to see, game on, what would happen,” said Root in the book. //

the Marines, being Marines, found several ways to bollix the AI and achieved a 100 percent success rate.

Two Marines, according to the book, somersaulted for 300 meters to approach the sensor. Another pair hid under a cardboard box.

“You could hear them giggling the whole time,” said Root in the book.

One Marine stripped a fir tree and held it in front of him as he approached the sensor. In the end, while the artificial intelligence knew how to identify a person walking, that was pretty much all it knew because that was all it had been modeled to detect. //

The moral of the story? Never bet against Marines, soldiers, or military folks in general. The American military rank-and-file has proven itself more creative than any other military in history. Whether that creativity is focused on finding and deleting bad guys or finding ways to screw with an AI and the eggheads who programmed it, my money's on the troops.

Last month, the Secretary of the Air Force put on a flight suit and sat in the front seat of an F-16.

His F-16 spent an hour in the air, dogfighting with another Air Force fighter. His jet was piloted by AI. //

I was reminded of the scene from "2001: A Space Odyssey." Machines deciding what is right and wrong. //

jumper
16 minutes ago edited
Between a president that keeps threatening to use F-15's against us and a woke military that will absolutely fire on their own people we may as well take our chances with the computers.

But the reality is quite different. This isn't "AI" in the sense that it's sentient and self-determinant. It's adaptive software that eliminates the problems of the human in the aircraft. There would be hard-wired kill switches and all sorts of other safety measures that sci-fi tries to pretend is easily bypassed. Put it this way: the Chinese and the Russians will be designing their own UCAV's. We would be foolish to fall behind in this.

Getting an AI to distinguish red from orange was a major challenge. //

The last time a human set the world record for solving a Rubik's Cube, it was Max Park, at 3.13 seconds for a standard 3×3×3 cube, set in June 2023. It is going to be very difficult for any human to pull off a John Henry-like usurping of the new machine record, which is more than 10 times faster, at 0.305 seconds. That's within the accepted time frame for human eye blinking, which averages out to one-third of a second.

TOKUFASTbot, built by Mitsubishi Electric, can actually pull off a solve in as little as 0.204 seconds on video, but not when Guinness World Records judges were measuring. The previous mechanical record was 0.38 seconds.

Billionaire Elon Musk said this month that while the development of AI had been “chip constrained” last year, the latest bottleneck to the cutting-edge technology was “electricity supply.” Those comments followed a warning by Amazon chief Andy Jassy this year that there was “not enough energy right now” to run new generative AI services. //

“One of the limitations of deploying [chips] in the new AI economy is going to be ... where do we build the data centers and how do we get the power,” said Daniel Golding, chief technology officer at Appleby Strategy Group and a former data center executive at Google. “At some point the reality of the [electricity] grid is going to get in the way of AI.” //

Such growth would require huge amounts of electricity, even if systems become more efficient. According to the International Energy Agency, the electricity consumed by data centers globally will more than double by 2026 to more than 1,000 terawatt hours, an amount roughly equivalent to what Japan consumes annually.

The promise of AI, we hear over and over again, is that it’s a tool to help humans do better, automating tasks to free up worker time for other things. But instead, AI looks far more like HAL 9000 in “2001: A Space Odyssey,” a computer that overtakes its human masters’ ability to control it and turns against humanity. //

Behind the scenes and out of sight, AI and social media algorithms can be used to determine what you are allowed to post, what you will be able to read, and ultimately what you will think.

Despite the promises of simplifying workflows and managing tasks, there’s far too much evidence of AI destruction to be ignored.

When it comes to AI, be afraid, be very afraid.

Swarovski AX Visio, billed as first "smart binoculars," names species and tracks location.

Last week, Austria-based Swarovski Optik introduced the AX Visio 10x32 binoculars, which the company says can identify over 9,000 species of birds and mammals using image recognition technology. The company is calling the product the world's first "smart binoculars," and they come with a hefty price tag—$4,799.