Apple using News to train AI models; AI companies required to disclose copyrighted training data

🤖 Google might already be replacing some human workers with AI. Link

• Google is considering the use of AI to "optimize" its workforce, potentially replacing human roles in its large customer sales unit with AI tools that automate tasks previously done by employees overseeing relationships with major advertisers.

• The company's Performance Max tool, enhanced with generative AI, now automates ad creation and placement across various platforms, reducing the need for human input and significantly increasing efficiency and profit margins.

• While the exact impact on Google's workforce is yet to be determined, a significant number of the 13,500 people devoted to sales work could be affected, with potential reassignments or layoffs expected to be announced in the near future.

🍎 Apple wants to use the news to help train its AI models. Link

• Apple is in talks with major publishers like Condé Nast and NBC News to license news archives for training its AI, with potential deals worth $50 million.

• Publishers show mixed reactions, concerned about legal liabilities from Apple's use of their content, while some are positive about the partnership.

• While Apple has been less noticeable in AI advancements compared to OpenAI and Google, it's actively investing in AI research, including improving Siri and other AI features for future iOS releases.

👺 Intel CEO says Nvidia was 'extremely lucky' to become the dominant force in AI. Link

• Intel CEO Pat Gelsinger suggests Nvidia's AI dominance is due to luck and Intel's inactivity, while highlighting past mistakes like canceling the Larrabee project as missed opportunities.

• Gelsinger aims to democratize AI at Intel with new strategies like neural processing units in CPUs and open-source software, intending to revitalize Intel's competitive edge.

• Nvidia's Bryan Catanzaro rebuts Gelsinger, attributing Nvidia's success to clear vision and execution rather than luck, emphasizing the strategic differences between the companies.

⚖️ AI companies would be required to disclose copyrighted training data under new bill. Link

• The AI Foundation Model Transparency Act requires foundation model creators to disclose their sources of training data to the FTC and align with NIST’s AI Risk Management Framework, among other reporting requirements.

• The legislation emphasizes training data transparency and includes provisions for AI developers to report on "red teaming" efforts, model limitations, and computational power used, addressing concerns about copyright, bias, and misinformation.

• The bill seeks to establish federal rules for AI transparency and is pending committee assignment and discussion amidst a busy election campaign season.

🚫 80% of Americans think presenting AI content as human-made should be illegal. Link

• According to a survey by the AI Policy Institute, 80% of Americans believe it should be illegal to present AI-generated content as human-made, reflecting broad concern over ethical implications in journalism and media.

• Despite Sports Illustrated's denial of using AI for content creation, the public's overwhelming disapproval suggests a significant demand for transparency and proper disclosure in AI-generated content.

• The survey also indicated strong bipartisan agreement on the ethical concerns and legal implications of using AI in media, with 84% considering the deceptive use of AI unethical and 80% supporting its illegalization.