openai, nvidia and google Updates

The artificial intelligence landscape is rapidly evolving, marked by significant investments and emerging challenges. OpenAI is reportedly in talks for a massive $100 billion deal with Nvidia for AI chips, but a critical hurdle is securing the immense power required to operate these advanced systems. The US power grid faces strain from the growing demand for AI data centers, which are projected to consume trillions of dollars in construction and hardware by 2029. This energy demand is so substantial that it rivals the power needs of entire cities, prompting exploration of solutions from gas turbines to renewable energy, though concerns about inflated energy forecasts and potential fossil fuel investments persist. Meanwhile, AI's integration into various sectors continues. Google is expanding its AI-powered conversational search experience, AI Mode, globally to Spanish-speaking users. In cybersecurity, Broadcom's Eric Chien advocates for machine learning tools to disrupt attackers, while Cloudflare's Christian Reilly advises a balanced approach to AI security, warning against both outright bans and uncontrolled adoption. The legal field is also grappling with AI's impact, as evidenced by a New Jersey attorney fined $3,000 for submitting fake AI-generated case law, highlighting the risks of unverified AI output. On the defense front, GE Aerospace and RTX are partnering with companies like Merlin and Shield AI, respectively, to integrate AI co-pilot and autonomous capabilities into military aircraft and drones. Educational institutions like Yale are seeing a surge in student interest in AI, with numerous AI-focused student groups forming. Additionally, smart contract security firm Sherlock has launched an AI auditor to enhance code review during development.

Key Takeaways

• OpenAI is reportedly negotiating a $100 billion deal with Nvidia for AI chips, facing significant challenges in securing the necessary electricity.
• Global spending on AI data centers is projected to reach $3 trillion by 2029, with immense power requirements posing a major infrastructure challenge.
• Concerns exist that AI energy demand forecasts may be inflated, potentially leading to unnecessary investment in fossil fuel projects.
• A New Jersey attorney was fined $3,000 for using AI to generate fake case law, underscoring the need for verification of AI-generated legal content.
• Google is expanding its AI-powered conversational search, AI Mode, globally to Spanish-speaking users.
• GE Aerospace is partnering with Merlin to develop AI co-pilot technology for military aircraft, aiming for autonomous flight capabilities.
• RTX and Shield AI have been selected by the Air Force to provide autonomous capabilities for Collaborative Combat Aircraft drone wingmen.
• Broadcom suggests using machine learning tools to disrupt cyber attackers, while Cloudflare advises calculated AI security measures.
• Student interest in AI is surging at universities like Yale, with increased activity in AI-focused student groups.
• Sherlock has launched an AI auditor to improve the security of smart contracts by identifying vulnerabilities during code development.

AI data centers require massive investment and power

The world will spend about $3 trillion on data centers for AI by 2029, with half going to construction and half to hardware. These centers need immense power, sometimes equivalent to thousands of homes using kettles at once. Companies are exploring solutions like gas turbines and renewable energy sources to meet this demand. The intense energy use also raises environmental concerns, with some regions considering water consumption limits for new sites. Despite high costs and challenges, AI data centers are seen as crucial for future technology.

OpenAI and Nvidia's $100B deal faces power access challenge

OpenAI has announced a $100 billion deal with Nvidia for AI chips, but securing enough electricity to power them is a major hurdle. The US power grid is struggling to keep up with data center construction, and adding OpenAI's demand would be like powering New York City at peak times. Utilities nationwide need to supply significant new power for data centers. This energy access issue is a critical bottleneck for AI development, forcing companies to explore creative solutions like building their own power plants.

AI energy demand forecasts may be inflated, risking fossil fuel projects

New reports suggest that projections for AI data center energy needs might be too high, potentially leading the US to invest in unnecessary and costly fossil fuel projects. While AI does consume significant electricity, the demand forecasts from utilities could be inflated due to speculative development and potential double-counting. This uncertainty risks higher utility bills and more pollution if new gas plants are built based on overestimated needs. Experts recommend more transparency and a focus on renewables like solar and wind to avoid these risks.

NJ lawyer fined $3,000 for using AI to create fake case law

A federal judge in New Jersey fined attorney Sukjin Henry Cho $3,000 for submitting fake case law generated by artificial intelligence in a court filing. The judge warned that lawyers using AI without verifying its output do so at their own risk. Cho admitted to using AI for legal research and blamed tight deadlines for the error, stating he has implemented stricter checks. This case is one of several recent instances where AI misuse in court has led to sanctions for attorneys.

NJ attorney fined $3,000 for AI-generated fake legal citations

Fort Lee attorney Sukjin Henry Cho was fined $3,000 by a federal judge for submitting fake case law created by artificial intelligence in a court filing. The judge noted that AI can generate realistic-sounding but fabricated legal propositions. Cho admitted to using generative AI tools for research and cited time constraints, promising to implement safeguards. This incident highlights a growing issue of AI misuse in courts, with other attorneys also facing fines for similar offenses.

GE Aerospace partners with Merlin for AI co-pilot technology

GE Aerospace is teaming up with Merlin to integrate AI into its avionics systems, starting with the Air Force's KC-135 tanker cockpit upgrade. This collaboration aims to develop an AI co-pilot that can assist human pilots, potentially reducing crew sizes on aircraft like the C-130J. Merlin's AI technology is designed to be adaptable and can process instructions from air traffic control automatically. While initially requiring human oversight, the long-term goal is for the AI to eventually fly planes autonomously.

RTX and Shield AI selected for Collaborative Combat Aircraft autonomy

The Air Force has chosen RTX and Shield AI to provide autonomous capabilities for its Collaborative Combat Aircraft (CCA) drone wingmen. RTX will supply the autonomy software for General Atomics' YFQ-42A, while Shield AI will provide it for Anduril's YFQ-44A. These selections are part of the first increment of the CCA program, which aims to develop advanced drone wingmen. While Anduril was not selected for the autonomy contract, it is still developing its aircraft.

Agile companies provide AI tools to operators

As the Pentagon accelerates innovation in artificial intelligence, agile companies are finding ways to provide AI tools to operators. This involves adapting government contracting processes to include non-traditional defense companies. RAFT, for example, offers AI products and data software designed to solve problems for operators. The focus is on how these companies can effectively deliver advanced AI solutions within the defense sector.

Yale sees surge in AI student groups and interest

Student-led organizations focused on artificial intelligence at Yale University are experiencing a significant increase in interest and membership. Groups like the Yale Artificial Intelligence Association, Yale Artificial Intelligence Alignment, and Yale Artificial Intelligence Policy Initiative offer various activities, from technical projects to discussions on AI ethics and long-term risks. This growth reflects broader debates about AI on college campuses and the university's own initiatives in AI research and policy.

Sherlock launches AI auditor to boost smart contract security

Sherlock, a smart contract security firm, has released a beta version of its AI auditor, Sherlock AI. This tool uses artificial intelligence to review code during development, aiming to identify vulnerabilities earlier than traditional audits. The goal is to complement existing security measures and provide faster, cheaper feedback to developers. This launch comes amid ongoing concerns about smart contract security and high-profile exploits in the crypto industry.

ASMSA hosts AI policy conference for gifted teachers

The Arkansas School for Mathematics, Sciences, and the Arts (ASMSA) recently hosted a conference for gifted and talented educators. The event focused on discussing the integration of artificial intelligence policies into K-12 classrooms for students. Alicia Cotabish from the University of Central Arkansas spoke about AI integration in education.

Broadcom's Eric Chien advocates ML tools to disrupt attackers

Eric Chien from Broadcom suggests that government agencies should use machine learning tools to disrupt cyber attackers. He highlights technologies like Broadcom's Symantec Adaptive Security, which uses machine learning to tailor security to an organization's environment and block unauthorized processes. Chien also introduced a tool called Incident Prediction, based on large language models, which can predict an attacker's next moves with high confidence, helping security teams proactively block threats.

Cloudflare's Reilly advises calculated AI security approaches

Christian Reilly, field CTO at Cloudflare, advises organizations to adopt calculated approaches to AI security amidst an 'AI gold rush.' He warns against extremes of banning AI or allowing uncontrolled use, emphasizing the need for risk management and clear guardrails. Reilly discussed challenges like 'shadow AI,' content scraping, and practical methods for enabling AI productivity without compromising security.

Google Search AI Mode expands globally in Spanish

Google Search is rolling out its AI Mode feature, an AI-powered conversational search experience, to Spanish-speaking users worldwide. This expansion follows recent global rollouts in multiple languages and introduces features like natural language queries and image uploads. Google also announced broader availability for its AI Plus subscription plan and new AI features for its Google AI Ultra subscribers, continuing its rapid integration of AI across its products.

Sources

• What's the big deal about AI data centres?
• The big challenge to OpenAI's $100B deal with Nvidia: Access to power
• The AI-energy apocalypse might be a little overblown
• What this N.J. lawyer did with AI landed him a hefty fine and a warning to all attorneys
• Bergen County lawyer fined $3,000 for misuse of artificial intelligence
• GE Aerospace picks Merlin for AI co-pilot, with eyes on KC-135 CCR upgrade [EXCLUSIVE]
• RTX, Shield AI picked to give Collaborative Combat Aircraft authonomous capabilities
• How agile companies are providing AI tools to operators
• AI student groups grow amid debates spanning clubs and classrooms
• Sherlock introduces AI auditor in beta to reinforce smart contract security
• WATCH | ASMSA holds artificial intelligence policy conference for gifted and talented teachers | Hot Springs Sentinel Record
• Eric Chien on Disrupting Attackers With ML-Based Tools
• AI 'Gold Rush' Demands Calculated Security Approaches
• Google’s AI Mode arrives in Spanish globally