Howdy! It’s Barsee again.

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

Today’s climb through the Valley reveals:

• Tesla achieves autonomous car delivery

• Runway is going to let people generate video games with AI

• AI models detecting Parkinson’s through a smile

• Facebook is starting to feed its AI with private, unpublished photos 📸🤖

• Plus trending AI tools, posts, and resources

Let’s dive into the Valley of AI…

FLOW

Stop typing. Start Flow-ing ✨

The fastest way to move work forward isn't a new shortcut—it's your own voice.

Wispr Flow now runs natively on Mac, Windows, and iPhone, so the same friction-free dictation follows you from the commute to your desk and back again.

Why teams are switching:

• 4× faster than typing. Speak at 140 wpm and let Flow handle the punctuation, re-phrasing, and formatting.

• Real-time AI edits & commands. Mumble, back-track, or change your mind, Flow cleans it up before you hit Send.

• Personal dictionary that travels with you. Learn names, acronyms, and industry jargon once, then sync them to every device.

• Tone awareness for every app. Sound crisp in Gmail, casual in Slack, and precise in VS Code without lifting a finger.

• 100+ auto-detected languages. Toggle effortlessly between English, Spanish, Polish, and more.

• Built for privacy & scale. SOC 2-ready and trusted by thousands of professionals who can't afford typos—or downtime.

One hotkey. Any workflow.

• Mac: Press Fn to start speaking in Notion, Figma, or anywhere the cursor blinks.

• Windows: Tap Ctrl + Win and dictate Jira updates, code comments, or long emails.

• iPhone: Whisper into the mic on the subway—Flow silences background noise and outputs publication-ready text.

Everything you say appears instantly across devices; your notes, drafts, and custom terms stay perfectly in sync.

Ready to Flow?

Grab the latest build for your platform of choice (or all three) and feel the speed bump on your very next message.

→ Download Flow free

(PS: Already using Flow on one device? Install it on the rest in minutes and watch your productivity compound.)

_{*This is sponsored}

THROUGH THE VALLEY

1. Tesla achieves autonomous car delivery from factory to customer

Tesla has reached a major achievement in its autonomous driving by delivering a fully driverless vehicle from the factory to a customer. In a video shared online, a Model Y can be seen traveling from Tesla’s Austin Gigafactory to a customer’s home without a human driver or operator. The Tesla was fully autonomous without a single human on board throughout the entire delivery process. This highlights the growing confidence in Tesla’s Full Self-Driving system, though its robotaxi service still requires human supervision.

2. Runway is going to let people generate video games with AI

Runway is launching a new tool called Game Worlds that lets users create basic video games using text prompts and uploaded images. Set to release next week, this marks the first time a text-to-game AI tool like this will be accessible to the general public. Initially, it will generate simple text-based adventures with visuals, with more complex game creation features expected later this year. Runway’s AI has already been used in major film and music projects, including Everything Everywhere All at Once, and its CEO says similar productivity boosts could soon help game developers speed up production.

3. AI models detecting Parkinson’s through a smile

A new AI breakthrough could make early Parkinson’s screening as simple as recording a smile. Researchers trained an AI model to detect the disease by analyzing short videos of people mimicking facial expressions, with nearly 88% accuracy across participants in the U.S. and Bangladesh. Published in the New England Journal of Medicine, the study shows the AI can catch subtle muscle changes caused by Parkinson’s (ones even doctors might miss). With no cure and limited access to neurologists, especially in rural areas, this tool could offer a low-cost, accessible solution for early diagnosis. The code is now open-source on GitHub.

4. UK entry-level jobs have dropped 32% since ChatGPT launch

Since ChatGPT’s launch in late 2022, entry-level job openings in the UK have dropped nearly 32%, with AI increasingly replacing roles once filled by new grads and junior staff, according to Adzuna. Entry-level jobs now make up just 25% of the UK job market, down from almost 29%. Companies like Klarna and IBM are leaning heavily on AI for tasks like customer service and HR, while leaders like Anthropic’s CEO warn that AI could wipe out half of all office jobs within five years. Meanwhile, AI-skilled workers are earning 56% more on average, and skill demands are shifting fast across many roles.

5. What LLMs know about their users

Simon Willison explored ChatGPT’s new memory dossier feature, revealing just how much the model (and OpenAI) can know about individual users. He shared a detailed JSON-formatted profile generated from his chat history, including sections like response preferences, past conversation highlights, and personal interests such as pelicans, cocktails, and database optimization. The memory showed deep insights, from where he lives to his coding habits and favorite topics. It’s a striking example of how LLMs not only store user context but also synthesize it into structured summaries. Willison calls it a “new world of intimate surveillance,” raising big questions about privacy and transparency.

6. Meta reportedly hires four more researchers from OpenAI

Meta has hired eight researchers from OpenAI in a recent talent grab, including four new names just reported: Shengjia Zhao, Jiahui Yu, Shuchao Bi, and Hongyu Ren. This follows earlier reports of other high-profile OpenAI departures to Meta, including Trapit Bansal. The hiring wave comes after Meta’s Llama 4 AI model reportedly underperformed, pushing the company to boost its AI efforts. OpenAI had downplayed Meta’s offers, but the talent loss has sparked concern. In response, OpenAI is now adjusting compensation and finding new ways to keep top researchers from jumping ship.

PEAK OF THE DAY

Facebook is starting to feed its AI with private, unpublished photos 📸🤖

Facebook is now asking users for access to their phone’s camera roll to suggest AI-edited versions of their photos, even those that haven't been uploaded yet.

Here's what you need to know:

• When creating a new Story on Facebook, you’ll see a prompt asking if you’d like to enable “cloud processing” for creative suggestions such as collages, AI edits, or themed photo ideas.

• By selecting “Allow,” you grant Facebook access to upload your photos to its cloud for analysis based on factors like time, location, and themes. The AI will then generate personalized suggestions, including collages or edits, which you can view privately before sharing. Importantly, Facebook assures that your photos won’t be used for targeted ads.

• However, enabling this feature means agreeing to Meta’s AI Terms of Service, which allows the company to analyze your photos (such as facial features, timestamps, and objects in the frame) to make creative suggestions.

• Although this feature is optional, it does raise privacy concerns. Your unpublished photos are also analyzed, which goes beyond Meta’s previous disclosure. You can opt out anytime through Facebook settings under “Camera roll sharing suggestions,” where you can toggle photo suggestions and cloud processing on or off.

• Currently, the feature is being tested in the U.S. and Canada. During this test phase, Meta assures users that photos uploaded for cloud processing will not be used for training AI models or for targeted ads.

Why it matters:

This feature is part of Meta’s broader AI push, tapping into your personal media before it's shared. It raises questions about how companies handle private data and the terms users are agreeing to when they give apps access to their personal information. Meta’s AI Terms state that once shared, your images may be analyzed by AI for personalization, and any personal data submitted could be used for future AI development. This could extend beyond what Meta originally revealed, bringing up new privacy concerns.

TRENDING TOOLS

• Ability AI Ads: From product link to viral Meta ad in 5 minutes. AI that powers your inner ad genius. *

• Deskminder: A simple yet effective app for people who get easily distracted. Create quick reminders & start tasks on your desktop.

• Underlord: An AI video editor for vibe editing.

• Workflows: A powerful new way to orchestrate complex legal tasks.

_{*asteric is sponsored}

THINK PIECES / BRAIN BOOST

• This 67-page AI report on how 300 execs at software startups like Cursor, ElevenLabs, and Sierra use AI has just dropped.

• It's known as ‘The List' and it's a secret file of AI geniuses by Mark Zuckerberg.

• How we accidentally solved robotics by watching 1 million hours of YouTube.

THE VALLEY GEMS

What’s trending on social today:

1/ OpenAI's upcoming open-source model is expected to outperform its competitors. It's set to launch next month.

2/ Neuralink announced their big plans for the coming years.

3/ Interesting things are happening.

4/ AI videos will be indistinguishable by 2027.

5/ When AI is built to work more like the human brain.

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”.