I think there's a strong argument for this, based on definition of TDM which is "analysis" and "analytical" only (which imply no downstream uses except "research") — nothing to do with reproducing inputs or automating labour. But it does go against the intuition of legislators.
ETHICS
-
Bot or Expert: Moral and Intellectual Consistency in AI Systems
By
–
I think Ollie/Jenny is a bot, but if not, there is nothing but my admiration for your work and expertise in our interactions here! You have got things consistently right, both morally and intellectually. If there's any hint of frustration it's with the system itself…
-
Human bias affects model’s count of R’s in strawberry
By
–
I’m just guessing this is the reason but the modal answer from all human statements about the count of R’s in “strawberry” might actually be 2. So it does it because we do.
-
OpenAI Launches ChatGPT Voice Mode After Safety Delays
By
–
In June, I reported that OpenAI delayed the launch of its new voice mode for ChatGPT to address safety issues; at the time, it said it would arrive in a month. Today @sama posted it'll drop next week (for a limited group of paid users). https://
bloomberg.com/news/articles/
2024-06-25/openai-delays-launch-of-voice-assistant-to-address-safety-issues?srnd=undefined
… -
Copyright Traps Paper Enhances Membership Inference Attacks Detection
By
–
Quoted in nice article by @Melissahei
, on a #ICML2024 paper on "copyright traps" by @matthieu_meeus @ffuuugor @ManuelFaysse @yvesalexandre Cool idea to enhance the efficacy of membership inference attacks to detect copyrighted data. Caution tho: similar to Glaze, not futureproof -
Fair Use Loopholes in AI Training Dataset Selection
By
–
I get that too. Fair Use and exceptions depend on "Use" and "purpose" — I hope they consider *all* of the uses not just a couple seeded by LAION that look innocent. (If not, that's another loophole!)
-
LAION Dataset and Copyright Infringement Legal Implications Analysis
By
–
The P2P cases like The Pirate Bay… downstream uses of files hosted elsewhere, the consequences of linking, are consistently part of the assessment. The LAION case has many of those aspects. A complaint about infringement due to dataset inclusion would have succeeded IMHO.
-
LLMs Imitate Training Data Format During Post-Training Stage
By
–
It does that because all of its training data in the last, post-training stage are of the form [question -> authoritative sounding solution], where the solutions are written by humans. The LLMs just imitate the form/style of that training data.
-
LLM Wrapper Papers Undermining Human-Computer Interaction Research
By
–
LLM Wrapper Papers are Hurting HCI Research https://
bit.ly/461w5y1
#AI #MachineLearning #DeepLearning #LLMs #DataScience -
Powerful AI Governance: International Coalition Against Authoritarian Control
By
–
the most powerful AI controlled by countries or groups that want to use it to cement their hold on power rather than to improve the lives of their citizens is a scary thing. a U.S.-led coalition of nations is the best way to prevent that.