South Africa recently withdrew its first AI policy framework after just 16 days due to the discovery of fake research citations. The policy had cited non-existent journal articles generated by AI, highlighting the risks of using AI without proper oversight. The Department of Communications and Digital Technologies must now address concerns about accountability, transparency, and explainability.
In a related development, a new report provides a framework for governments to assess their readiness for agentic AI, which can coordinate entire workflows and deliver outcomes across organizational boundaries. The framework maps 70 core government functions against the potential for agentic AI to add public value and the complexity of deployment.
Meta, a key player in the AI sector, has seen a deal reversal in its acquisition of Manus, deepening the split between China and Silicon Valley. Chinese start-ups are now focusing on domestic markets, while US investors are cautious due to regulatory scrutiny. This reflects a fragmenting landscape in the tech sector.
On a more practical note, a new repository called smol-audio offers self-contained Jupyter notebooks for fine-tuning AI models like Whisper, Parakeet, and Voxtral. These notebooks are designed for use in Google Colab and aim to bridge the gap between AI model development and practical application.
Former IRS Commissioner Daniel Werfel emphasizes the need for careful risk management in AI adoption, focusing on delivery over compliance. Two chefs, Meenu Bhasin and Melanie Underwood, are using AI for operational tasks such as researching ingredients and planning events, highlighting the potential for AI to augment human capabilities.
Meta's AI glasses have also enabled a veteran who lost his eyesight 30 years ago to 'see' again, using AI to help him perceive his surroundings. Meanwhile, Michigan lawmakers are considering a proposal for a statewide AI pilot program to test AI in a controlled government setting.
Key Takeaways
- South Africa withdrew its AI policy framework due to fake research citations generated by AI.- A new report provides a framework for governments to assess readiness for agentic AI.
- Meta's deal reversal deepens the split between China and Silicon Valley.
- smol-audio offers Jupyter notebooks for fine-tuning AI models like Whisper and Parakeet.
- Former IRS Commissioner emphasizes careful risk management in AI adoption.
- Chefs use AI for operational tasks, highlighting its potential to augment human capabilities.
- Meta's AI glasses help a veteran 'see' again.
- Michigan lawmakers consider a statewide AI pilot program.
- RecursiveMAS enables more efficient agent collaboration via unified latent-space recursive computation.
- Agentic AI can coordinate workflows and deliver outcomes across organizational boundaries.
South Africa's AI policy fiasco: fake research cited
South Africa's first attempt at an AI policy framework was withdrawn after just 16 days due to fake research citations. The policy was found to have cited non-existent journal articles generated by AI. This incident highlights the risks of using AI without proper oversight. The Department of Communications and Digital Technologies must now answer questions about accountability, transparency, and explainability. The policy's failure demonstrates the need for human checks and robust governance in AI use.
South Africa withdraws AI policy due to fake citations
South Africa's draft national AI policy was withdrawn after it was discovered that at least six of its 67 academic citations were AI-generated hallucinations. The policy aimed to position the country as an AI innovation leader and included plans for new institutions to oversee AI use. The incident highlights the growing issue of AI-generated fake references and the need for careful human oversight.
A practical guide to implementing agentic AI in government
A new report provides a framework for governments to assess their readiness for agentic AI. The framework helps governments prioritize and sequence AI deployment in a responsible and effective manner. Agentic AI can coordinate entire workflows and deliver outcomes across organizational boundaries. The report maps 70 core government functions against the potential for agentic AI to add public value and the complexity of deployment.
Meta Deal Reversal Deepens Split Between China and Silicon Valley
Meta's acquisition of Manus has been reversed, deepening the split between China and Silicon Valley. Chinese start-ups are now raising money at home and building for domestic markets, while US investors steer clear of the scrutiny that comes with backing them. This reflects a fragmenting landscape in the tech sector.
smol-audio: A Colab-Friendly Notebook Collection for Fine-Tuning AI Models
smol-audio is a repository of self-contained Jupyter notebooks for fine-tuning AI models like Whisper, Parakeet, and Voxtral. The notebooks are designed to be used in Google Colab and provide practical knowledge for working with these models. smol-audio aims to close the gap between the development of AI models and their practical application.
Scaling Agent Collaboration via Recursion
RecursiveMAS is a new approach to scaling agent collaboration via unified latent-space recursive computation. This approach enables more efficient and effective collaboration among agents.
Ex-IRS Chief: AI adoption needs careful risk management
Former IRS Commissioner Daniel Werfel emphasizes the need for a strategic shift in AI adoption, focusing on delivery over compliance and managing risks. He highlights the importance of careful risk management in AI implementation.
Two chefs use AI for operational help in solo cooking companies
Two chefs, Meenu Bhasin and Melanie Underwood, use AI to help with operational tasks such as researching ingredients, planning events, and developing social strategy. They emphasize that AI cannot replace the human touch in teaching and tasting.
New Meta AI glasses help veteran 'see' the room around him
A veteran who lost his eyesight 30 years ago can now 'see' again thanks to a pair of Meta AI glasses. The glasses use AI to help him perceive his surroundings.
Michigan lawmakers consider statewide AI pilot program
Michigan lawmakers are considering a proposal to test AI in a controlled government pilot program. The program would allow agencies to experiment with generative AI tools under strict oversight.
Sources
- South Africa’s AI policy cited fake research, created by AI: what lessons need to be learned
- South Africa withdraws AI policy after it was found written by AI
- How governments can make agentic AI re ?
- Meta Deal Reversal Deepens Split Between China and Silicon Valley
- smol-audio: A Colab-Friendly Notebook Collection for Fine-Tuning Whisper, Parakeet, Voxtral, Granite Speech, and Audio Flamingo 3
- Scaling Agent Collaboration via Recursion
- Ex-IRS Chief: AI adoption needs careful risk management
- Two chefs running solo cooking companies tap AI for operational help
- New Meta AI glasses help veteran ‘see’ the room around him
- Michigan lawmakers consider statewide AI pilot program
Comments
Please log in to post a comment.