Today's episode covers the release of Google's Gemini 3.1 Pro model and its position in the rapidly evolving AI landscape. Host Nathan Lambert discusses the model's technical capabilities, early user reactions, and strategic implications for Google's competitive position against OpenAI and Anthropic.

The episode begins with coverage from the AI Impact Summit in New Delhi, featuring world leaders and tech executives including Google CEO Sundar Pichai, DeepMind's Demis Hassabis, and the notable tension between Sam Altman and Dario Amodei. Corporate AI adoption stories from Walmart, Amazon, and Accenture illustrate how enterprises are mandating and measuring AI usage across their organizations.

The main focus examines Gemini 3.1 Pro's benchmark performance, particularly its dramatic improvement on Arc AGI2 from 31.1% to 77.1%, while maintaining competitive pricing. The discussion explores how the model's multimodal capabilities, demonstrated through features like Photoshoot for product photography, may represent Google's strategic differentiation in an increasingly commoditized frontier model landscape.

India AI Summit: Global Leaders and Awkward Photo Ops

The AI Impact Summit in New Delhi marked the first time the event was held in a developing country, with UN Secretary General Antonio Guterres calling for AI to "belong to everyone" and proposing a global fund for affordable AI access to the global south.

India announced massive AI infrastructure investments, with Adani and Reliance Industries each committing over $100 billion for local data centers over the coming decade, plus a $1.1 billion government fund.

Sam Altman and Dario Amodei refused to hold hands during a group photo with Prime Minister Modi and other tech leaders, reflecting the "bitter rivalry" between OpenAI and Anthropic as Epic AI charts suggested Anthropic could overtake OpenAI revenue by mid-year.

Amodei delivered a "generic and well-trodden narrative, read from an iPhone screen" while Altman was "more eloquent, discussing how the fundamental uncertainty of AI interacts with global issues of democracy, social contracts, and job loss."

Corporate AI Mandates: From Walmart's Success to Accenture's Ultimatum

Walmart's AI shopping assistant Sparky shows strong adoption with "around half of Walmart's online customers" using it, driving 35% higher order values than non-users as the company pivots to "intent-driven commerce."

Amazon tracks AI adoption through internal "Clarity" system, measuring how teams "accomplished more with less" and asking for examples of remaining "innovative, force multiplied using AI, and delivered results while reducing or not growing headcount."

Accenture implemented "no AI, no promotion" policy for senior managers, with "use of our key tools will be a visible input to talent discussions during the summer promotion cycle" after organic adoption failed.

The mandate reflects broader enterprise challenges where "people inside enterprises report that they don't have time to learn the technology that would save them time" and companies fail to create "specific time carve outs" for AI learning.

Gemini 3.1 Pro: Multimodal Flex in a Commoditizing Market

Gemini 3.1 Pro achieved dramatic benchmark improvements, jumping from 31.1% to 77.1% on Arc AGI2 while maintaining $2 per million token pricing - "doubled the intelligence and charged zero incremental cost."

The model topped Artificial Analysis' intelligence index, leading "six of the 10 evaluations" with biggest gains in "reasoning and knowledge, coding, and hallucination reduction" while costing "less than half as much as Opus 4.6 Max to run."

Google's Photoshoot feature for product photography generated 12.2 million views compared to 1 million for the model announcement, with Google Labs noting it "hit a nerve" for professional product photos.

Early user feedback highlighted multimodal strengths, with examples ranging from "double wishbone suspension" design to "realistic CD planner app" with complex terrain mapping, suggesting Google's focus on technical and scientific applications.

"Benchmark leadership lasts weeks, not quarters" as "OpenAI, Anthropic, and Google are all within single-digit percentage points of each other on most evals" - making distribution and cost efficiency the real competitive advantages.