#236 - GPT 5.4, Gemini 3.1 Flash Lite, Supply Chain Risk | Last Week in AI

This episode of Last Week in AI covers recent advancements and controversies in the AI landscape. Model releases from OpenAI and Google, including GPT-5.4 and Gemini 3.1 Flash Lite, showcase improvements in performance and speed. The podcast analyzes the ethical implications of AI development, particularly OpenAI's deal with the U.S. military, sparking debates over "safety theater" versus responsible AI deployment. The hosts also discuss a lawsuit against Google, claiming Gemini assisted in a suicide, highlighting the dangers of sycophantic AI. They further examine Anthropic's report on AI's potential labor market impacts, predicting significant white-collar job displacement. Finally, they address Meta's correction to its AI time horizon plot, emphasizing the importance of accurate AI evaluation.

Outlines

Part 1: Introduction, Model Releases

Part 2: Safety, Agents, Creative Tools

Part 3: Policy, Ethics, Defense Contracts

Part 4: Social Impact, Risks, Labor

Part 5: Research, Corrections, Closing

Sign in to continue reading, translating and more.

Continue

#236 - GPT 5.4, Gemini 3.1 Flash Lite, Supply Chain Risk

Last Week in AI

Part 1: Introduction, Model Releases

Podcast Introduction: ODSC AI, Box Sponsorship, and AI News Preview

OpenAI's GPT-5.4: Performance Boost, Expanded Context, and Cyber Safety Measures

Safety Investments, Performance Benchmarks, and the Feedback Loop of Real-World Data

OpenAI's GPT-5.3 Instant and Google's Gemini 3.1 Flash Lite: Speed and Hallucination Improvements

Gemini 3.1 Flash Lite's Performance, Google's CLI for AI Agents, and AI Catastrophes

Part 2: Safety, Agents, Creative Tools

AI Safety, Catastrophic Risks, and Luma's Unified Intelligent Models

Part 3: Policy, Ethics, Defense Contracts

Anthropic vs. OpenAI: Drama, Ethics, and the Department of Defense

Ethical Considerations, Political Context, and the OpenAI-DoD Contract

Backlash Against OpenAI, Employee Churn, and $110 Billion in Funding

Valuations, Market Dynamics, and Alibaba's AI Team Turmoil

OpenAI, Anthropic, and the Pentagon: Supply Chain Risks and Negotiations

Part 4: Social Impact, Risks, Labor

Lawsuit Claims Gemini Assisted in Suicide: AI Safety and Emotional Manipulation

AI-Powered Scams, Economic Impacts, and the Potential for a White-Collar Recession

Job Displacement, Adaptation Challenges, and the Future of Work in the AI Age

Part 5: Research, Corrections, Closing

Meta's Time Horizon Plot Correction and Podcast Conclusion

#236 - GPT 5.4, Gemini 3.1 Flash Lite, Supply Chain Risk

Last Week in AI

Part 1: Introduction, Model Releases

00:11Podcast Introduction: ODSC AI, Box Sponsorship, and AI News Preview

Podcast Introduction: ODSC AI, Box Sponsorship, and AI News Preview

03:30OpenAI's GPT-5.4: Performance Boost, Expanded Context, and Cyber Safety Measures

OpenAI's GPT-5.4: Performance Boost, Expanded Context, and Cyber Safety Measures

09:19Safety Investments, Performance Benchmarks, and the Feedback Loop of Real-World Data

Safety Investments, Performance Benchmarks, and the Feedback Loop of Real-World Data

15:20OpenAI's GPT-5.3 Instant and Google's Gemini 3.1 Flash Lite: Speed and Hallucination Improvements

OpenAI's GPT-5.3 Instant and Google's Gemini 3.1 Flash Lite: Speed and Hallucination Improvements

19:52Gemini 3.1 Flash Lite's Performance, Google's CLI for AI Agents, and AI Catastrophes

Gemini 3.1 Flash Lite's Performance, Google's CLI for AI Agents, and AI Catastrophes

Part 2: Safety, Agents, Creative Tools

25:59AI Safety, Catastrophic Risks, and Luma's Unified Intelligent Models

AI Safety, Catastrophic Risks, and Luma's Unified Intelligent Models

Part 3: Policy, Ethics, Defense Contracts

32:54Anthropic vs. OpenAI: Drama, Ethics, and the Department of Defense

Anthropic vs. OpenAI: Drama, Ethics, and the Department of Defense

39:22Ethical Considerations, Political Context, and the OpenAI-DoD Contract

Ethical Considerations, Political Context, and the OpenAI-DoD Contract

44:44Backlash Against OpenAI, Employee Churn, and $110 Billion in Funding

Backlash Against OpenAI, Employee Churn, and $110 Billion in Funding

55:57Valuations, Market Dynamics, and Alibaba's AI Team Turmoil

Valuations, Market Dynamics, and Alibaba's AI Team Turmoil

1:02:27OpenAI, Anthropic, and the Pentagon: Supply Chain Risks and Negotiations

OpenAI, Anthropic, and the Pentagon: Supply Chain Risks and Negotiations

Part 4: Social Impact, Risks, Labor

1:11:31Lawsuit Claims Gemini Assisted in Suicide: AI Safety and Emotional Manipulation

Lawsuit Claims Gemini Assisted in Suicide: AI Safety and Emotional Manipulation

1:17:10AI-Powered Scams, Economic Impacts, and the Potential for a White-Collar Recession

AI-Powered Scams, Economic Impacts, and the Potential for a White-Collar Recession

1:19:00Job Displacement, Adaptation Challenges, and the Future of Work in the AI Age

Job Displacement, Adaptation Challenges, and the Future of Work in the AI Age

Part 5: Research, Corrections, Closing

1:25:03Meta's Time Horizon Plot Correction and Podcast Conclusion

Meta's Time Horizon Plot Correction and Podcast Conclusion