Welcome to AI Tool Report!

Thursday’s top story: Ex-Google CEO has been in the headlines this week, for slamming Google’s choice to prioritize employees’ work-life balance over beating its rivals, something he has now, unexpectedly retracted.

🌤️ This Morning on AI Tool Report

1. 😮 Ex-Google CEO’s shock U-turn

2. 💻 How to get the world’s largest models working on local devices

3. 🚀 Li’s start-up hits $1B in 4 months!

4. 📧 How to improve email subscriber growth using ChatGPT

5. 🚨 First ever AI risk database

6. 😱 Shocking truth about AI models!

Read Time: 5 minutes

STOCK MARKETS

✂️ AI stocks had a day of consolidation after increasing steadily over the past week and a half. Investors seem to be waiting for more information on rate cuts before recession fears can be put to rest. Learn more.

 — — — — — —

POLICIES

😮 Ex-Google CEO’s shock U-turn

Our Report: After ex-Google CEO—Eric Schmidt—hit the headlines earlier this week for slamming Google for prioritizing ‘working from home’ over beating rivals, like OpenAI, and for telling impressionable students at Stanford University (in a video interview) that it was ok for AI startups to steal IP and then just hire a lawyer to handle legal disputes, he has now retracted one of his bombshell statements.

🔑 Key Points:

• Schmidt–who was Google’s CEO for 10 years—said that Google was putting itself at a disadvantage by deciding that "work-life balance and working from home was more important” than beating rivals.

• He also said that in Silicon Valley, AI start-ups steal IP and hire lawyers to deal with the aftermath because “if the product’s a hit, they just need a lawyer to solve the legal problems; if the product fails, the theft is insignificant.”

• He has since retracted one of his comments, regretfully admitting he "misspoke about Google and their work hours," and has asked Stanford University to delete the interview from YouTube, which they have.

🤔 Why you should care: Despite the backlash Schmidt faced over his comments about Google’s ‘‘detrimental’’ working-from-home policy, several industry leaders have previously agreed with him, with OpenAI’s CEO, Sam Altman, saying he believed that one of the industry's "worst mistakes" was allowing employees to go "full remote forever," Meta’s CEO, Mark Zuckerberg, believing that "people who are working from home are not efficient”, and Tesla’s CEO, Elon Musk, calling remote working “morally wrong.”

 — — — — — —

Together with WebAI

The world’s largest models.
On your own devices.

At our recent Summer Release Event, we announced three new capabilities:

• Navigator: Deploy your models directly from the canvas to your local device, or across a local network of devices working together.


• webFrame: Fine-tune large state-of-the-art models and run them across local devices, bringing AI directly to where the data is being created.


• Assistant: Tap into your models without connection to the outside world. Your data remains on your device and can run on your laptop, mobile, or even VR hardware.

Learn More

1. Meco is a distraction-free space for reading and discovering newsletters, separate from the inbox

2. Languify is an AI-powered web extension that optimizes copywriting

3. AI Copywriter uses AI to enhance landing page copy

4. LearnGPT enhances learning experiences using AI

5. AI Face Swap uses AI to swap faces in photos

— — — — — —

FUNDING

🚀 Li’s start-up hits $1B in 4 months!

Our Report: Esteemed Stanford University professor and “Godmother of AI”, Fei-Fei Li, has officially raised another $100M for her start-up—World Labs—which was led by Venture Capital firm, NEA, in the second round of funding since the company’s inception in April.

🔑 Key Points:

• Li initially raised $100M (led by a16z and Candian VC firm, Radical Ventures, where Li is a scientific partner) in May, and this latest funding has now propelled Wold Labs' valuation to over $1B (up from $200M in April).

• Li founded World Labs to develop AI models that process visual, 3D data with human-like understanding, enabling them to “do more than see and talk,” but also “do,” revolutionizing gaming, autonomous vehicles, and robotics.

• These models will have advanced reasoning and be able to understand the dimensions of objects, where things are, and what they do, which has the potential to open up new frontiers in healthcare and urban planning.

🤔 Why you should care: World Labs has gone from start-up to Unicorn (which are start-ups that are valued at over $1B) in just four months, thanks to these two funding rounds, highlighting that—despite the reported scarcity of start-up funding—investors believe in the potential of World Labs, helped of course by Li’s reputation as a groundbreaking AI scientist.

— — — — — —

Thursday’s Prompt: How to improve email subscriber growth using ChatGPT

Type this prompt into ChatGPT:

Results: After typing this prompt, you will get an in-depth analysis of why your email subscribers are churning and insight into how to increase subscriber growth.

P.S. Use the Prompt Engineer GPT by AI Tool report to 10x your prompts.

AI-Powered Video for Every Platform

Boost your video content game with Vizard! This AI tool converts a single video into 30+ platform-specific clips with one button. 

Ideal for TikTok, Instagram, and YouTube Shorts, Vizard provides AI-driven features like auto-reframing, speaker focus, and dynamic captions, ensuring your videos are always engaging and on-brand.

Want to streamline your video production? Subscribe to The AI Report for the complete story and exclusive insights!

— — — — — —

RISK

🚨 First ever AI risk database

• Massachusetts Institute of Technology (MIT) has created a database, documenting over 700 AI risks, to give researchers, developers, businesses, and policymakers "an accessible overview of the AI risk landscape.”

• MIT created the database after census data highlighted that "even the most thorough framework overlooks approximately 30% of the risks," so without a database, risks will be missed or not given proper consideration.

• The first-of-its-kind database includes risks ranging from discrimination and misrepresentation to fraud, targeted manipulation, and unsafe use, and MIT will ensure it’s “consistently updated” so it stays active.

— — — — — —

CHATBOTS

😱 Shocking truth about AI models!

• Researchers have created a benchmark for chatbot hallucinations by fact-checking models (including GPT-4o, Llama 3, Gemini 1.5 Pro, and Claude 3 Opus) on topics ranging from law and health to pop culture and celebrities.

• They found that “the greatest models generate hallucination-free text just 35% of the time” meaning, despite AI companies insisting the contrary, AI models can't be trusted and are hallucinating as much as they were. 

• Existing benchmarks that establish the “factuality” of AI models predominantly ask the model questions where the answers are on Wikipedia, while over ½ the answers in this test couldn’t be found on Wikipedia.

📺 Deepfake Ryan Gosling explains how ChatGPT works

We read your emails, comments, and poll replies daily.

Hit reply and tell us what you want more of!

Until next time, Martin & Liam.

P.S. Don’t forget, you can unsubscribe if you don’t want us to land in your inbox anymore.