Why Chatbots Still Hallucinate – and How OpenAI Wants to Fix It

In today’s email:

🍿 OpenAI Backs AI-Made Animated Feature Film, “Critterz”
🔥 GPT-5 Thinking in ChatGPT (aka Research Goblin) is shockingly good at search
🤑 Anthropic to Pay $1.5 Billion to Settle Book Piracy Class Action Lawsuit
🧰 13 new AI-powered tools and resources. Make sure to check the online version for the full list of tools.

OpenAI: Chatbots Hallucinate Because We Trained Them to Bluff

Key Takeaway: Hallucinations persist because current evals reward confident guessing and penalize uncertainty, so OpenAI proposes changing evaluation methods to incentivize abstaining when unsure.

More Insights:

Errors aren’t random—LLMs act like “permanent exam takers” optimized to guess rather than admit doubt.
Accuracy-only scoreboards encourage bluffing; silence/hedging gets treated as failure.
OpenAI contrasts Claude’s cautious style: fewer wrong claims, but more refusals that can reduce utility.
Proposed fix: update primary evals to stop penalizing abstentions and to reward calibrated uncertainty.
Shift in focus: from fluency and speed to reliability and humility, especially for high-stakes use.

Why it matters: Incentives shape behavior—if benchmarks prize lucky guesses, models will keep bluffing; redesigning evals could realign the whole ecosystem toward trustworthy, calibrated AI that knows when not to answer.

How 433 Investors Unlocked 400X Return Potential

Institutional investors back startups to unlock outsized returns. Regular investors have to wait. But not anymore. Thanks to regulatory updates, some companies are doing things differently.

Take Revolut. In 2016, 433 regular people invested an average of $2,730. Today? They got a 400X buyout offer from the company, as Revolut’s valuation increased 89,900% in the same timeframe.

No wonder 10K+ everyday people are taking the chance on Pacaso.

Founded by a former Zillow exec, Pacaso’s co-ownership tech reshapes the $1.3T vacation home market. They’ve earned $110M+ in gross profit to date, including 41% YoY growth in 2024 alone. They even reserved the Nasdaq ticker PCSO.

The same institutional investors behind Uber, Venmo, and eBay backed Pacaso. And you can join them. But not for long. Pacaso’s investment opportunity ends September 18.

Invest While You Still Can

_{Paid advertisement for Pacaso’s Regulation A offering. Read the offering circular at}_{invest.pacaso.com}_{. Reserving a ticker symbol is not a guarantee that the company will go public. Listing on the NASDAQ is subject to approvals.}

AI robots can already carve stone statues. Entire buildings are next

Key Takeaway: Monumental Labs is scaling AI-driven robotic stone carving from statues to structural blocks—aiming to make stone buildings fast, affordable, and low-carbon.

More Insights:

$8M round led by Seven Seven Six (Alexis Ohanian) funds a 37,000-sq-ft Brooklyn factory with 30-ft ceilings and a fleet of seven-axis carving robots.
Robots rough-cut; artisans hand-finish—already delivering restorations for Carnegie Hall and the Frick in weeks instead of months.
Next frontier: “structural stone” blocks for full facades and walls—targeting as little as 25% of concrete’s cost with automation.
In-house AI (reinforcement learning) will optimize toolpaths and cuts, aiming to cut fabrication costs by 80–90%.
A 30-ft stone observation tower inside the new facility will demo readiness; goal capacity: ~100 life-size sculptures/year.

Why it matters: If they crack automated structural stone, construction could shift from disposable glass-and-concrete to durable, lower-carbon masonry—reviving craftsmanship, retooling trades, and reshaping city skylines for centuries instead of decades.

OpenAI Backs AI-Made Animated Feature Film, “Critterz”

Key Takeaway: OpenAI is lending its tools and compute to “Critterz,” an AI-heavy animated film racing to finish in ~9 months for a Cannes debut and global theatrical release next year.

More Insights:

Budget under $30M—far below typical animated features—aiming to prove AI can slash costs and timelines.
Produced by Vertigo Films and Native Foreign; funded by Federation Studios, with profit-sharing for ~30 crew members.
Workflow blends human artists and voice actors with OpenAI models (including GPT-5) to turn sketches into final imagery.
Script contributions from members of the “Paddington in Peru” writing team; production has begun, casting soon.
No distributor yet; industry is wary amid copyright fights (e.g., lawsuits against Midjourney) and union concerns.

Why it matters: If “Critterz” lands with audiences, it could redefine animation economics—compressing multi-year pipelines into months—while forcing Hollywood, labor groups, and regulators to quickly re-draw the lines between human creativity, AI assistance, and copyright.

Other stuff

GPT-5 Thinking in ChatGPT (aka Research Goblin) is shockingly good at search
OpenAI and Retro Biosciences achieve 50x increase in expressing stem cell reprogramming markers.
OpenAI set to start mass production of its own AI chips with Broadcom
Anthropic to Pay $1.5 Billion to Settle Book Piracy Class Action Lawsuit
Inside the Anthropic ‘red team’ tasked with breaking its AI models
Sam Altman says that bots are making social media feel ‘fake’
I used AI to make this song. The result and its implications startled me
Computer scientist Geoffrey Hinton: ‘AI will make a few people much richer and most people poorer’

Win 1 year free Superpower ChatGPT Pro🤑

All your ChatGPT images in one place 🎉

You can now search for images, see their prompts, and download all images in one place.

Get the extension for FREE

Editor's Pick

Frontegg - AI agents without guardrails create chaos. Join our webinar to learn how to secure access. 👉 Save your spot

CapCut AI Suite - Create, edit, or remix content with AI in a simple editor

CTV ads made easy: Black Friday edition

As with any digital ad campaign, the important thing is to reach streaming audiences who will convert. Roku’s self-service Ads Manager stands ready with powerful segmentation and targeting — plus creative upscaling tools that transform existing assets into CTV-ready video ads. Bonus: we’re gifting you $5K in ad credits when you spend your first $5K on Roku Ads Manager. Just sign up and use code GET5K. Terms apply.

Use code GET5K now

Snipman - Ai-powered dynamic snippet manager