• AI Valley
  • Posts
  • $2,000 AI-generated ad on NBA finals

$2,000 AI-generated ad on NBA finals

PLUS: Alexandr Wang leaves Scale AI to lead Meta’s Superintelligence

Together with

Howdy! It’s Barsee again.

Happy Monday, AI family, and welcome back to AI Valley.

Today’s climb through the Valley reveals:

  • $2,000 fully AI-generated ad that aired during the NBA finals

  • Alexandr Wang leaves Scale AI to lead Meta’s Superintelligence

  • Trump administration's whole-government AI plans leaked on GitHub

  • Scientists confirm AI can achieve humanlike cognition

  • Plus trending AI tools, posts, and resources

Let’s dive into the Valley of AI…

WISPR FLOW

Image: Flow

Flow is the AI voice keyboard that turns speech into polished text in any iPhone app—Slack, iMessage, Gmail, Notion—5× faster than typing.

  • AI-polished output: removes fillers, auto-punctuates

  • Works everywhere: universal keyboard inside every app

  • Learns your lingo: remembers names & acronyms

  • Whisper freely: dictate quietly without disturbing others

  • Desktop sync: notes mirror to Mac & Windows

Free plan with weekly word cap; upgrade to Pro for unlimited dictation.

*This is sponsored

PEAK OF THE DAY

$2,000 fully AI-generated ad that aired during the NBA finals 🤖🏀

Credit: PJ Ace

In a groundbreaking moment for advertising and artificial intelligence, the 2025 NBA Finals featured the first-ever fully AI-generated commercial aired during a major primetime sports event.

Here's what you need to know:

  • Directed by self-described “AI filmmaker” PJ Accetturo, the ad was produced entirely using Google’s Veo 3, a text-to-video AI generator.

  • He utilized AI tools like Gemini and ChatGPT to develop the script, shot list, and prompts, resulting in 15 usable clips from 300–400 generations.

  • The entire production was completed in just two days at a cost of approximately $2,000, a stark contrast to traditional commercial production budgets that often reach six to seven figures.

  • Created for the prediction market platform Kalshi, used surreal and unconventional imagery, including a farmer floating in a pool of eggs, an alien chugging beer, and an elderly man draped in an American flag exclaiming, “Indiana gonna win baby.”

Why it matters:

As AI technology continues to evolve, the Kalshi ad serves as a compelling example of how artificial intelligence is going to revolutionize content creation, offering new possibilities for storytelling and audience engagement in the advertising industry.

AI video is now good enough for primetime ads, offering brands a way to create eye-catching (if surreal) content at a fraction of traditional costs. While some will still pay for "premium" human-crafted spots, the rise of tools like Veo 3 means weird, hyper-engaging AI ads will dominate 2025.

THROUGH THE VALLEY

Image Credit: TheInformation

Alexandr Wang is stepping down as CEO of Scale AI to join Meta and lead its superintelligence efforts, following Meta’s massive $14.3 billion investment in the startup. Although Meta now holds a 49% non-voting stake in Scale AI, the company says this won’t affect existing clients like Google, Microsoft, or OpenAI. Jason Droege, former Uber VP and current chief strategy officer, will take over as CEO. A few Scale employees will also move to Meta. This move reflects Mark Zuckerberg’s focus on advancing Meta’s AI capabilities after the lukewarm reception of its latest Llama model.

The Trump administration is quietly preparing to launch a new government-wide AI initiative on July 4, based on a now-hidden GitHub repository tied to the “ai.gov” website. Spearheaded by former Tesla engineer Thomas Shedd, the project aims to make the U.S. government operate like a tech startup, using AI to automate federal work. The site will reportedly include a chatbot, an “all-in-one API” for connecting to OpenAI, Google, and Anthropic models, and a dashboard called CONSOLE to monitor usage. While vendors are mostly FedRAMP certified, concerns remain over data security and rapid AI integration without clear oversight.

TikTok is expanding its AI-powered Symphony ad platform to let advertisers generate influencer-style videos using virtual avatars that model clothing, hold products, and showcase apps. Brands can upload images and text prompts to create polished promotional videos without relying on real creators. This automation reduces costs and increases content output, posing a threat to human influencers whose rates and relevance may decline. While TikTok promises safety reviews and AI-generated content labels, the move signals a shift toward scalable, synthetic advertising, where virtual avatars increasingly replace the traditional roles once held by real people in digital marketing.

As Big Tech pours billions into large-language models, top researchers like Fei-Fei Li and Yann LeCun are focused on building “world models” (AI systems that mimic how humans mentally represent and predict the real world). Unlike LLMs that rely on word patterns, world models aim to develop spatial intelligence and common sense by interacting with 3D environments. Li’s World Labs, backed by $230M, is pioneering this effort, while LeCun’s team at Meta is training AI on video data to simulate abstract understanding. Though challenged by limited spatial data, world models may be the true path to achieving human-level AI.

Chinese researchers have shown that large language models (LLMs) like ChatGPT and Gemini can spontaneously form humanlike ways of understanding and categorizing real-world objects, without being specifically trained to do so. In a study published in Nature Machine Intelligence, scientists tested models using “odd-one-out” tasks across 1,800+ objects and found that their conceptual groupings mirrored core aspects of human cognition. Multimodal models aligned even more closely with brain activity patterns, suggesting they process objects similarly to humans. While not identical, these models demonstrate emerging cognitive structures, paving the way for more intuitive AI-human interactions and future intelligent robots.

Image: Google

Google is testing a new feature called “Audio Overviews” that brings AI-generated podcast-style summaries to mobile search results. When enabled via Google Labs, users searching questions like “How do noise cancellation headphones work?” can tap a “Generate Audio Overview” button, triggering a 40-second process that creates a short discussion between two AI hosts. The audio appears in a mini player on the search page, with controls and links to source material. Currently limited to English in the U.S., this feature expands Google's efforts to offer audio responses in Search, building on tools like NotebookLM and Gemini.

TRENDING TOOLS

  • Taplio’s AI is trained on 500 M+ LinkedIn posts to help you write high-performing content in seconds. *

  • Copilot Vision is an AI companion that can see your PC in real time.

  • Huntr, a job search tool that automates the job search for you.

* asterisk is sponsored content

THINK PIECES / BRAIN BOOST

THE VALLEY GEMS

What’s trending on social today:

THAT’S ALL FOR TODAY

Thank you for reading today’s edition. That’s all for today’s issue.

💡 Help me get better and suggest new ideas at [email protected] or @heyBarsee

👍️ New reader? Subscribe here

Thanks for being here.

REACH 100K+ READERS

Acquire new customers and drive revenue by partnering with us

Sponsor AI Valley and reach over 100,000+ entrepreneurs, founders, software engineers, investors, etc.

If you’re interested in sponsoring us, email [email protected] with the subject “AI Valley Ads”.