This suggests a number of potential problems. For example, who vets the training data?
https://www.bespacific.com/a-i-s-un-learning-problem/
A.I.’s un-learning problem
Fortune: – Researchers say it’s virtually impossible to make an A.I. model ‘forget’ the things it learns from private user data …“If a machine learning-based system has been trained on data, the only way to retroactively remove a portion of that data is by re-training the algorithms from scratch,” Anasse Bari, an A.I. expert and computer science professor at New York University, told Fortune. The problem goes beyond private data. If an A.I. model is discovered to have gleaned biased or toxic data, say from racist social media posts, weeding out the bad data will be tricky. Training or retraining an A.I. model is expensive. This is particularly true for the ultra-large “foundation models” that are currently powering the boom in generative A.I. Sam Altman, the CEO of OpenAI, has reportedly said that GPT-4, the large language model that powers its premium version of ChatGPT, cost in excess of $100 million to train. That’s why, to companies developing A.I. models, a powerful tool that the U.S. Federal Trade Commission has to punish companies it finds have violated U.S. trade laws is scary. The tool is called “algorithmic disgorgement.” It’s a legal process that penalizes the law-breaking company by forcing it to delete an offending A.I. model in its entirety. The FTC has only used that power a handful of times, typically directed at companies who have misused data. One well known case where the FTC did use this power is against a company called Everalbum, which trained a facial recognition system using people’s biometric data without their permission…”
(Related)
The Case For Large Language Model Optimism in Legal Research From A Law & Technology Librarian
Via LLRX – The Case For Large Language Model Optimism in Legal Research From A Law & Technology Librarian – The emergence of Large Language Models (LLMs) in legal research signifies a transformative shift. This article by Sean Harrington critically evaluates the advent and fine-tuning of Law-Specific LLMs, such as those offered by Casetext, Westlaw, and Lexis. Unlike generalized models, these specialized LLMs draw from databases enriched with authoritative legal resources, ensuring accuracy and relevance. Harrington highlights the importance of advanced prompting techniques and the innovative utilization of embeddings and vector databases, which enable semantic searching, a critical aspect in retrieving nuanced legal information. Furthermore, the article addresses the ‘Black Box Problem’ and explores remedies for transparency. It also discusses the potential of crowdsourcing secondary materials as a means to democratize legal knowledge. In conclusion, this article emphasizes that Law-Specific LLMs, with proper development and ethical considerations, can revolutionize legal research and practice, while calling for active engagement from the legal community in shaping this emerging technology.
Still not generic enough?
https://www.brookings.edu/articles/a-comprehensive-and-distributed-approach-to-ai-regulation/
A comprehensive and distributed approach to AI regulation
While algorithmic systems have become widely used for many impactful socioeconomic determinations, these algorithms are unique to their circumstances. This challenge warrants an approach to governing algorithms that comprehensively enables application-specific oversight. To address this challenge, this paper proposes granting two new authorities for key regulatory agencies: (1) administrative subpoena authority for algorithmic investigations, and (2) rulemaking authority for especially impactful algorithms within federal agencies’ existing regulatory purview. This approach requires the creation of a new regulatory instrument, introduced here as the Critical Algorithmic Systems Classification, or CASC. The CASC enables a comprehensive approach to developing application-specific rules for algorithmic systems and, in doing so, maintains longstanding consumer and civil rights protections without necessitating a parallel oversight regime for algorithmic systems.
The accompanying illustration is brilliant!
How worried should you be about AI disrupting elections?
Disinformation will become easier to produce, but it matters less than you might think
All that is not forbidden is mandatory. All that is not mandatory is forbidden.
https://www.zdnet.com/article/one-in-four-workers-fears-being-considered-lazy-if-they-use-ai-tools/
One in four workers fears being considered 'lazy' if they use AI tools
Fear of being judged appears to be a major obstacle preventing workers from using AI.
No comments:
Post a Comment