- What's brewing in AI
- Posts
- The Sunday recap✨
Howdy, wizards.
⏪ This is your weekly recap email on Sundays. All the best links I’ve shared during the week. No fluff or unnecessary details included.
🤷🏻♂️ Don’t need the Sunday recap? You can update your subscriber preferences at the end of this email to pick what emails you’d like to receive.
Let’s recap!
THE SUNDAY RECAP
THE MOST IMPORTANT DEVELOPMENTS IN AI THIS WEEK
Screenshot from a16z.com // DALL-E
Venture capital firm Andreesen Horowitz (a16z) has released their yearly list of big ideas—predictions on what will be trending in tech—for 2025. The list features 50 of trends across 7 verticals and, naturally, there’s a lot of AI-related stuff in there.
Here’s a few of the key upcoming trends in AI:
Personalised AI-powered health monitoring and insights for everyone
AI alleviating staff shortages in healthcare by automating routine tasks
Real-time AI enabling new use cases that lets users prototype, iterate, and refine ideas on the fly
Building an AI brain – Turning AI into an infinite memory bank to help us make informed decisions.
Analyzing (and making use of) qualitative data becomes possible at scale with LLMs, web-based agents, and multimodal models
AI-native storytelling gets serious, enabling real-time, player-driven experiences, blurring the lines between film and gaming
AI companions get way more lifelike, having their own “inner worlds”, forming deeper relationships with users.
Google’s dominance will loosen, as more people use AIs like ChatGPT and Perplexity for general search, and domain-specific search engines like Consensus (academic research), Harvey (law), Hebbia (financial services) and more.
GenAI will transform sales, automate many current parts of sales reps’ work, allowing them to focus on high-touch, consultative selling.
a16z’s article might seem long, but it’s actually a super-condensed read to help you understand a lot about how tech will evolve in the near-term future. An actual treasure trove.
→ For all the AI-related trends (and my favourites), check out the full newsletter here
TOGETHER WITH TLDR
TLDR covers the best tech, startup, and coding stories in a quick email that takes 5 minutes to read. And it's read by over 1.2 million people!
OpenAI is doing a 12-day “ship-mas” where they’ll announce a new product or demo something every day over the next 12 weekdays.
On Friday, the first day, they introduced ChatGPT Pro and the full o1 model. Here’s what you need to know:
OpenAI’s most powerful model, o1, is officially out of preview. The full model will be available for all paid users (Plus and Team users from today, Edu and Enterprise next week). Key upgrades:
It’s a faster, more powerful model than 01-preview; it’s particularly better at coding, math and writing. They’re citing a 34% error-reduction on real-world questions.
Supports image uploads, ie you can upload photos & docs and have ChatGPT reason over them.
The report on the safety work done on the model, its system card, shows it maintains a “medium” risk.
Support for o1 in the API, including support for function calling, developer messages, structured outputs and vision is coming (but no date on this yet).
They’ve also launched a ChatGPT Pro subscription at a whooping $200/month. This includes a “pro mode” of the o1 model that’s even better than the regular o1, as well as unlimited access to all models and Advanced Voice.
The pro mode of o1 uses more compute to provide the best possible answer to the hardest problems (think difficult math and programming problems). Responses take more time so they have a progress bar and sends you in-app notification when they’re done.
Pro will be OpenAI’s tier where they’ll launch advanced capabilities that require a lot of compute to perform.
The Plus tier at $20/month isn’t going anywhere and is still the best choice for “almost everyone” according to Sam Altman.
Also, here’s what happens when you ask o1-pro to figure out a way to make money with the minimum amount of work.
I ran some tests on the full o1 for a complex prompt I’ve been using o1-preview for lately that includes categorising and summarising articles. The response time is indeed faster, and the writing feels materially less AI-like. I’m excited about it!
→ Read the full newsletter here
World Labs has unveiled a tool that generates interactive 3D scenes from a single photo. It creates explorable and modifiable environments right in your browser. The generated scenes come with interactive effects like dynamic lighting and adjustable depth of field.
The project is still in an early preview phase, and World Labs plans to improve the size and fidelity of these worlds and is aiming to have its first product ready for the market in 2025. They’ve already raised $230 million in funding and are targeting applications in gaming, movie production, and design.
For a cool preview, try to “step into” these paintings from Van Gogh, Hopper and more.
We’ve seen AI tools like generative fill and AI videos where you can extend clips or fill in parts that didn’t exist before. But this takes it a step further by turning a single image into an interactive 3D world.
→ Read the full newsletter here
OpenAI CEO Sam Altman spoke at New York Times’ DealBook Summit earlier this week. Here’s the most interesting highlights from the interview:
ChatGPT now has 300 million weekly active users and receives more than 1 billion messages per day. 1.3 million developers in the US are using OpenAI’s products.
Sam acknowledges that some tension exists in their relationship to Microsoft, but the two companies are overall aligned on incentives.
Most safety concerns related to AI won’t actually happen at the moment AGI is here, but on the long continuation towards true superintelligence. “We’ll hit AGI sooner than most people in the world think, and it will matter much less”.
Sam sees deep learning as sort of like a law of physics, and thinks AI will historically be viewed as something that was first discovered by a few companies, but over time will become commoditised – and that’s fine since science should be diffused throughout society. He says the AI will be used for everything and that it will be inconceivable to people in the future that their devices are not all smart.
When asked how he feels, on a personal level, about Elon Musk suing OpenAI, he replied “tremendously sad”, explaining that he grew up with Elon as a hero, followed by praise for Elon for making the world think more ambitiously, himself included. Altman’s perception of the lawsuit, “whatever the stated claim is”, is that Elon sees OpenAI as a competitor, and is resorting to legal action to weaken them, rather than simply competing in the marketplace.
→ Read the full newsletter here
DARIO’S PICKS
The Verge just dropped an immersive, high-quality piece on AI companions. Companionship is so far one of the biggest use cases for AI – just look at the raging success of character.ai, or the market map a16z did earlier this year.
Not only is the user base for such companions probably bigger than most people realise, it’s also a lot more diverse. And it brings up interesting questions on the ethics of AI, and insights on the direction and speed at which things are moving.
Beyond the productivity gains AI will have, our relationship to the technology in itself is going to be transformative. Obviously, we’re not all going to have an AI spouse anytime soon, but looking at these fringes cases of users—and the extent to which they’re anthropomorphizing LLMs—can tell us something about what relationship the mainstream will have with AI, and help identify the ways in which AI companies are contributing to what could be an impending, collective mass-delusion by making their agents increasingly human-like.
Reading this article made me more aware of my own relationship with AI – and the subtle, unconscious ways I treat and think about ChatGPT/Claude/other AIs as human. I don’t think there’s necessarily anything bad about it, but bringing awareness into our thoughts and actions is always a good idea.
→ For the most interesting snippets from the article, read the full newsletter here
THAT’S ALL FOLKS!
 | Was this email forwarded to you? Sign up here. Want to get in front of 13,000 AI enthusiasts? Work with me. This newsletter is written & curated by Dario Chincha. |
What's your verdict on today's email? |
Affiliate disclosure: To cover the cost of my email software and the time I spend writing this newsletter, I sometimes link to products and other newsletters. Please assume these are affiliate links. If you choose to subscribe to a newsletter or buy a product through any of my links then THANK YOU – it will make it possible for me to continue to do this.