Forgot your password?
typodupeerror

Submission + - SPAM: Should Open Source AI Mean Exposing All Training Data?

kazekiri writes: We have examined what constitutes the “preferred form of making modifications” for AI in the philosophical, legal, and technical contexts. Philosophically, granting freedom for all components that affect an AI model’s performance is admirabpractical interpretation of many jurisdictions is that any rights in the training data do not extend to the resulting model. Coupled with privacy restrictions on certain datasets, requiring complete data disclosure can clash with reality. Meanwhile, from a technical angle, the code’s algorithm and pipeline are often more critical in defining how the model behaves, and the actual need for full data to achieve near-equivalent reproduction is limited.

Bringing this together suggests that mandating full dataset release as a requirement for “preferred form of making modifications” is not necessarily realistic. Instead, adequate documentation of how others might assemble or locate similar data can suffice to maintain alignment with existing laws and social norms. Although a purely philosophical approach to openness might champion complete training data, OSI’s approach—requiring training code, parameters, and comprehensive Data Information—represents a pragmatic balance that encourages broader adoption of Open Source AI.

Link to Original Source

Submission + - DeepSeek in China: A Legal Overview of the Generative AI Regulation (shujisado.org)

kazekiri writes: In 2023, China enacted the Provisional Measures on the Management of Generative Artificial Intelligence Services, which serves as the regulatory guidance for the generative AI services industry.
China’s AI regulations have two seemingly contradictory aspects: on the one hand, they establish a strong censorship system, and on the other hand, they emphasize international cooperation for technological development. This reflects China’s strategic approach to striking a balance between technological development and security.

Comment Hidden Traps in Meta’s Llama License (Score 1) 35

There is an article explaining that not only is Llama not open source, but it is also a license with a very high risk.
Meta can update the AUP at any time, and it is possible to trap all users. In addition, this trap is also effective for users of services that use Llama.

https://shujisado.org/2025/01/...

Social Networks

Ask Slashdot: Do We Need Pseudonymous Social Networking? 213

An anonymous reader writes "While the idea of anonymous social networking sounds like an oxymoron, the use of pseudonyms to mask a user's online identity has a long history that stretches back to the earliest days of the Internet and local bulletin board systems (BBS). Such imperfect anonymity, which can often be unmasked with a few well-defined Google searches, has led to abuses like the invention of 'spambots' and the persistence of forum trolls. But, as the BBC reports, pseudonyms have their place in online communities, especially where identities are a risky commodity, under oppressive state regimes and governments where corporate interests increasingly dominate the interests of individuals: 'Some users choose to hide their identity to avoid being found by people they would not like to be contacted by. Others live in countries where identification could have serious implications for those who have expressed political views or associated themselves with others who have.' Should Google+ and maybe even the notorious Facebook evolve into two-tiered sites where those who choose to remain anonymous are 'identified' as such and denied access to certain site features, while being free to post, blog, or tweet their views, without summarily getting their accounts suspended or revoked?"
User Journal

Journal Journal: . Tea Party Hypocrites 10

Well, well, well. This one's from my own state (even though folks in Chicago think I-80 is Illinois' southern state line). Another "fiscal conservative", a teabagger who demands the US balance its budget and even demands a Constitutional amendment who can't even balance his own budget. The AP

Earth

Today's Lighter TVs Mean Much Less E-Waste 197

MojoKid writes "We all know that today's flat-screen TVs weigh far less than old-style CRTs, or they wouldn't be able to hang on the wall. New research from the Consumer Electronics Association finds that this translates into a massive savings of electronics waste. The report found that today's flat screen TVs are 82% lighter and 75% smaller than cathode ray tube (CRT) TVs. In other words, 40- to 70-inch flat-panel TVs weigh 34% less than 13- to 36-inch CRT TVs. This reduction in materials has a staggering downstream effect. The report claimed that an old 36-inch CRT TV generated about the same amount of electronics waste as 5,080 cell phones. However, today's 70-inch flat-screen TV generate the equivalent of just 953 cell phones."
Businesses

Jimmy Wales' Theory of Failure 164

Hugh Pickens writes "The Tampa Tribune reports that Jimmy Wales recently spoke at the TEDx conference in Tampa about the three big failures he had before he started Wikipedia, and what he learned from them. In 1996 Wales started an Internet service to connect downtown lunchers with area restaurants. 'The result was failure,' says Wales. 'In 1996, restaurant owners looked at me like I was from Mars.' Next Wales started a search engine company called 3Apes. In three months, it was taken over by Chinese hackers and the project failed. Third was an online encyclopedia called Nupedia, a free encyclopedia created by paid experts. Wales spent $250,000 for writers to make 12 articles, and it failed. Finally, Wales had a 'really dumb idea,' a free encyclopedia written by anyone who wanted to contribute. That became Wikipedia, which is now one of the top 10 most-popular Web sites in the world. This leads to Wales' theories of failure: fail faster — if a project is doomed, shut it down quickly; don't tie your ego to any one project — if it stumbles, you'll be unable to move forward; real entrepreneurs fail; fail a lot but enjoy yourself along the way; if you handle these things well, 'you will succeed.'"
Space

This Is the Way the World Ends 394

Dave Knott writes "The CBC's weekly science radio show Quirks and Quarks this week features a countdown of the top ten planetary doomsday scenarios. Nine science professors and one science fiction author are asked to give (mostly) realistic hypotheses of the ways in which the planet Earth and its inhabitants can be destroyed. These possibilities for mankind's extinction include super-volcanoes, massive gamma ray bursts, and everybody's favorite, the killer asteroid. Perhaps the most terrifying prediction is the reversal of the Earth's magnetic field (combined with untimely solar activity), a periodic event which is currently 1/4 million years overdue."
Medicine

Saving 28,000 Lives a Year 263

The New Yorker is running a piece by Atul Gawande that starts by describing the everyday miracles that can be achieved in a modern medical intensive care unit, and ends by making a case for a simple and inexpensive way to save 28,000 lives per year in US ICUs, at a one-time cost of a few million dollars. This medical miracle is the checklist. Gawande details how modern medicine has spiraled into complexity beyond any person's ability to track — and nowhere more so than in the ICU. "A decade ago, Israeli scientists published a study in which engineers observed patient care in ICUs for twenty-four-hour stretches. They found that the average patient required a hundred and seventy-eight individual actions per day, ranging from administering a drug to suctioning the lungs, and every one of them posed risks. Remarkably, the nurses and doctors were observed to make an error in just one per cent of these actions — but that still amounted to an average of two errors a day with every patient. Intensive care succeeds only when we hold the odds of doing harm low enough for the odds of doing good to prevail. This is hard." The article goes on to profile a doctor named Peter Pronovost, who has extensively studied the ability of the simplest of complexity tamers — the checklist — to save lives in the ICU setting. Pronovost oversaw the introduction of checklists in the ICUs in hospitals across Michigan, and the result was a thousand lives saved in a year. That would translate to 28,000 per year if scaled nationwide, and Pronovost estimates the cost of doing that at $3 million.

Slashdot Top Deals

Per buck you get more computing action with the small computer. -- R.W. Hamming

Working...