June 7, 2025

news  
Read News
AI News: Veterans Affairs, Google, MIT, and Ray-Ban

This news brief covers a range of recent developments in the field of Artificial Intelligence. A ProPublica report revealed that the Department of Government Efficiency (DOGE) used a flawed AI model, created by engineer Sahil Lavingia, to review contracts at the Department of Veterans Affairs, resulting in misdirected cuts to critical veterans' care, including cancer treatment research. The AI, which was intended to find waste, only reviewed the initial parts of contracts and lacked necessary context, leading to the cancellation of contracts not directly supporting patient care and redirecting $900 million back to the agency. Google has shared insights into how user behavior has shaped the development of its AI Mode, noting that users desire more predictable AI Overviews capable of handling longer questions. However, Google's AI Overviews have also been criticized for spreading false information, reducing website traffic, and providing inaccurate summaries, such as advising users to add glue to pizza sauce. In other AI advancements, MIT researchers have launched Boltz-2, an open-source AI model for drug discovery that is faster and more accurate than previous models, and Oro Labs has introduced an AI Agent Builder to its procurement platform, enabling procurement teams to create AI agents for tasks like risk reviews and fraud detection. An AI security camera has been developed to protect vehicles from theft by detecting suspicious activities, while in the hospitality sector, AI is being used to personalize customer experiences and automate tasks, though experts emphasize the irreplaceable value of human interaction. In the realm of AI policy, Congress is considering legislation that could preempt state AI safeguards for 10 years, potentially impacting state-level protections for children and workers. The Philippines is leading in generative AI course enrollments, demonstrating a strong commitment to AI skills development, with the government aiming to train 1 million AI-skilled workers by 2028. Finally, the Ray-Ban Meta AI glasses have been criticized for their design which incorporates non-replaceable batteries, leading to obsolescence and contributing to e-waste.

Key Takeaways

  • DOGE's flawed AI led to misdirected cuts in Veteran Affairs contracts, negatively impacting veterans' care.
  • Google's AI Overviews are facing criticism for spreading false information and reducing website traffic.
  • MIT researchers released Boltz-2, an open-source AI model for drug discovery.
  • Oro Labs added an AI Agent Builder to its procurement platform for tasks like risk reviews and fraud detection.
  • An AI security camera can detect suspicious activity around vehicles to prevent theft.
  • AI is being used in hospitality to personalize experiences, but human touch remains crucial.
  • Federal legislation could override state AI safeguards, impacting protections for children and workers.
  • The Philippines leads in generative AI course enrollments, showing a strong commitment to AI skills.
  • Ray-Ban Meta AI glasses are designed with non-replaceable batteries, contributing to e-waste.
  • Google is using user behavior to improve and shape the development of its AI Mode.

ProPublica reveals how DOGE's AI cuts hurt veterans' care

ProPublica reports that DOGE used flawed AI to review VA contracts, leading to critical cuts. Engineer Sahil Lavingia, who worked with DOGE, admitted the AI code had flaws. The AI was supposed to find waste, but instead, it cut contracts for things like cancer treatment research. Experts say the AI only looked at part of the contracts and didn't understand important details. This resulted in poor decisions that negatively impacted veterans' care.

Report: DOGE's AI tool wrongly cut Veteran Affairs contracts

The Department of Government Efficiency (DOGE) used a flawed AI model to review contracts at the Department of Veterans Affairs. Engineer Sahil Lavingia created the AI, which canceled contracts not directly supporting patient care. The AI only reviewed the first part of the contracts and lacked clear definitions, leading to mistakes. The VA initially planned to cancel 875 contracts, but reduced the number to 585 after concerns were raised. The cuts redirected $900 million back to the agency.

Report: DOGE's AI tool wrongly cut Veteran Affairs contracts

The Department of Government Efficiency (DOGE) used a flawed AI model to review contracts at the Department of Veterans Affairs. Engineer Sahil Lavingia created the AI, which canceled contracts not directly supporting patient care. The AI only reviewed the first part of the contracts and lacked clear definitions, leading to mistakes. The VA initially planned to cancel 875 contracts, but reduced the number to 585 after concerns were raised. The cuts redirected $900 million back to the agency.

Google reveals how user behavior shaped AI Mode development

Google shared details on how user behavior influenced the creation of AI Mode. User research showed people wanted AI Overviews to be more predictable and handle longer questions. Google identified key use cases like exploratory advice, how-to guides, and local shopping. Experts have raised concerns about traffic loss for websites due to AI Overviews. Google says AI search delivers more qualified clicks, but has no data to share. The company plans to roll out more AI Mode features soon.

Google's AI Overviews spread false and dangerous information

Google's AI Overviews are reportedly giving wrong information and hurting publishers. The AI tool summarizes search results but sometimes makes up facts, called hallucinations. For example, it advised adding glue to pizza sauce. Click-through rates to websites drop significantly when AI Overviews are shown. Google says it is working to improve the tool. Google CEO Sundar Pichai claims the tool helps users find a wider range of sources.

Boltz-2 democratizes AI molecular modeling for drug discovery

MIT researchers released Boltz-2, an open-source AI model for drug discovery. Boltz-2 predicts how well molecules will bind together, which helps in finding new drugs. It is faster and more accurate than previous models. Unlike AlphaFold 3, Boltz-2 is open-source, allowing more researchers to use it. Recursion, an AI drug discovery company, collaborated on the project. Boltz-2 helps understand how drugs work and can speed up drug development.

AI security camera protects vehicles from theft

Vehicle thefts are increasing, and current security measures are not enough. An AI security camera can monitor vehicles and detect suspicious activity. The camera uses computer vision to recognize actions like tire tampering or window breaking. The Arduino Portenta H7 with a Vision Shield can run the AI model. The Edge Impulse platform helps train and deploy the model. The system can send alerts when potential theft is detected.

Oro Labs adds AI agent builder to procurement platform

Oro Labs has added an AI Agent Builder to its procurement platform. This allows procurement teams to create and use their own AI agents. The AI agents can handle tasks like risk reviews, fraud detection, and compliance. The platform includes pre-built AI agents for supplier maintenance and knowledge retrieval. Companies using Oro's AI have reduced processing time by 50%. The AI Agent Builder has a drag-and-drop interface for easy creation and customization.

AI can't replace human touch in hospitality

AI is impacting the hospitality industry in many ways. AI systems analyze customer data to personalize recommendations and marketing. They also automate HR tasks and provide virtual assistants for guests. Robots are used for tasks like check-in, cleaning, and food delivery. However, experts believe AI cannot replace the human touch in hospitality. Smart hotel rooms use AI to improve efficiency and guest satisfaction, but personal interaction remains important.

Federal AI power grab could hurt state protections for kids, workers

Congress is considering a move that could stop states from enforcing AI safeguards. Section 43201 would override state laws governing AI for 10 years. Supporters say it's needed to avoid a mix of state rules that could hurt U.S. AI competitiveness. Critics say it could undermine state efforts to protect children and workers. States have been leading the way in online safety and data privacy. The moratorium could block state privacy laws and protections for children on social media.

Philippines leads in GenAI course enrollments, beating US and China

The Philippines is leading the world in generative AI (GenAI) course enrollments. Enrollments in the Philippines surged by 383%, surpassing the global average. This shows the country's strong commitment to AI. The Philippine workforce is becoming more confident in its digital skills. Filipinos are using AI to improve skills and assist with writing tasks. The government aims to train 1 million AI-skilled workers by 2028.

Ray-Ban Meta AI glasses designed for the dump

The Ray-Ban Meta AI glasses have sleek designs but a troubling flaw they are designed to become obsolete. The glasses have non-replaceable batteries, limiting their lifespan. When the battery dies, the entire product becomes useless. Manufacturing these glasses requires mining precious metals and produces carbon emissions. E-waste is increasing rapidly, and these glasses contribute to the problem. Sustainable alternatives with replaceable batteries are possible, but Meta's design promotes waste.

Sources

AI Artificial Intelligence DOGE Department of Veterans Affairs VA contracts AI flaws Sahil Lavingia Google AI Overviews Hallucinations Click-through rates Boltz-2 Drug discovery Open-source AI model MIT AlphaFold 3 AI security camera Vehicle theft Computer vision Arduino Portenta H7 Edge Impulse Oro Labs AI Agent Builder Procurement platform Hospitality AI in hospitality AI safeguards State laws AI competitiveness Data privacy Generative AI GenAI Philippines Ray-Ban Meta AI glasses E-waste Sustainability AI Mode User behavior AI molecular modeling