AI Highlights

My top-3 picks of AI news this week.

xAI
1. Grok goes free-for-all

Musks’ xAI has made a bold move by making their AI assistant Grok freely available to all X users, alongside significant improvements and new features.

• Democratising access: Grok is now available for free to all X users, not just premium subscribers.

• Enhanced capabilities: The new version is 3x faster with improved accuracy, instruction-following, and multilingual support, while adding web search and citations.

• Aurora integration: Launched a new image generation model called Aurora, enabling photorealistic image creation and profile picture reimagining through the “draw me” feature.

Alex’s take: This move fascinates me because it challenges the traditional AI business model. Whilst others are focusing on premium pricing (cough, OpenAI…), xAI is betting on widespread adoption through X's platform. It almost feels how Google made search free two decades ago—a strategy that fundamentally changed how we access information.

Google
2. From Genesis to Gemini

Google has unveiled Gemini 2.0 highlighting a serious evolution in their AI capabilities, focusing on agentic workflows.

• Multimodal mastery: Beyond just understanding multiple inputs (text, video, images, audio, code), Gemini 2.0 can now generate multimodal outputs including images and multilingual text-to-speech.

• Native tool integration: The model can natively interface with Google Search, code execution, and third-party functions, making it more capable of completing complex tasks.

• Project showcase: Introduced several groundbreaking projects, including Project Astra (universal AI assistant), Project Mariner (browser-based agent), and Jules (GitHub-integrated coding assistant).

Alex’s take: While previous AI models excelled at understanding and generating content, Gemini 2.0 is designed to actively navigate and interact with digital environments. The ability to control browsers, manage GitHub workflows, and even play games suggests that we’re moving from an era of passive AI assistants to proactive AI agents.

OpenAI
3. OpenAI unwrapped

Just when we thought OpenAI might slow down for the holidays, they've dropped even more updates as part of their “12 Days of OpenAI” special:

• Sora: Their new text-to-video model capable of generating cinema-quality videos up to 20 seconds long.

• Canvas: A new collaborative workspace in ChatGPT for drafting and editing, with direct Python code execution and GPT integration.

• Advanced Voice: Real-time video chat and screen sharing capabilities rolling out to ChatGPT mobile app users.

• Apple Integration: Deep OS-level integration allowing ChatGPT to work seamlessly with Siri, visual intelligence, and composition tools.

Alex’s take: OpenAI is systematically breaking down barriers between human and AI interaction. However, not all barriers are technological. Being based in the UK, I still can't try Sora due to ever-growing regulations. Innovation-first regions race (US) ahead while regulation-first regions fall behind (EU). It’s been fascinating to see movements like eu/acc emerge to change this.

Content I Enjoyed

Zuck and Musk join forces against OpenAI

As if the Sam Altman saga wasn't enough drama for one year, OpenAI is back in the spotlight.

This time, we're witnessing an unlikely alliance as Meta urges California's attorney general to block OpenAI's planned conversion to a for-profit company, effectively joining forces with Elon Musk, who filed a similar request in November.

Something that stood out to me was Meta's warning in their letter to Attorney General Rob Bonta about the “seismic implications for Silicon Valley.”

If OpenAI's transition is allowed, it could set a precedent where startups enjoy nonprofit benefits until they're ready to profit, while investors benefit from both tax write-offs and for-profit returns.

Meanwhile, OpenAI's chairman Bret Taylor assures that any restructuring would ensure the nonprofit “continues to exist and thrive.”

But here's where it gets interesting: OpenAI just released emails showing Musk himself advocated for a for-profit structure back in 2015, highlighting that “salary from the nonprofit muddies the alignment of incentives.”

As OpenAI stated: “We have great respect for Elon’s accomplishments and gratitude for his early contributions to OpenAI, but he should be competing in the marketplace rather than the courtroom.”

I have to say, it is interesting to see the game theory behind situations like this. Musk launched his own AI company, xAI, in July 2023. Let’s see how this one plays out.

Idea I Learned

A quantum leap that's bigger than the universe itself

This week, I explored Google's announcement of their new quantum chip, Willow, and learned something mind-bending.

Willow completed a calculation in 5 minutes that would take today's most powerful supercomputer 10 septillion years. For perspective, that's longer than the universe has existed.

The breakthrough lies in their approach to error correction. Traditionally, quantum computers face a fundamental problem where more qubits (quantum bits) lead to more errors.

Through error correction, Willow achieves something remarkable: adding more qubits reduces errors.

This has been a holy grail in quantum computing since 1995, and Google just cracked it.

While we're still waiting for the first commercially relevant quantum application, this breakthrough feels like one of those pivotal moments in computing history—like the first transistor or the first microprocessor.

As Alan Turing, the father of modern computing, once said: “We can only see a short distance ahead, but we can see plenty there that needs to be done.”

We're all witnessing the early days of a technology that will likely reshape our world in ways we can barely imagine.

I thought this video was a great primer on what error correction means for quantum computing.

Quote to Share

Deedy Das on AI video generation’s current limits:

“As cool as the new Sora is, gymnastics is still very much the Turing test for AI video.”

This observation comes as OpenAI's new text-to-video model Sora struggles with generating realistic gymnastic movements, producing hilariously distorted body shapes that look more like abstract art than human athletics.

The comment taps into something deeper about AI development. Complex physical activities expose the current limitations of AI's understanding of human biomechanics.

Much like the infamous “Will Smith eating spaghetti” video that became a benchmark for early video-generation models, it looks like gymnastic movements have emerged as an unexpected stress test for video AI.

Source: @deedydas on X

Question to Ponder

“Give me a lever long enough and a place to stand, and I will move the earth.” — Archimedes.

Over the last few decades, one Silicon Valley brag has followed a rather corrosive trend.

“The larger the headcount of my company, the more successful I am perceived amongst my peers.”

The reality is that this may, more often than not, be a negative signal.

You see, I’m a big believer that technology ultimately boils down to two things. Scale and leverage. Yet the Valley has been historically accelerating in the opposite direction.

Throwing people at a problem instead of leveraging technology and outsourcing to scale.

Auren Hoffman highlights this perfectly:

“If you start stripping out everything that is not unique to your company, you’re left with just a few people who make the unique parts of the company. And then add a few people who need to explain its unique benefits to the market. Imagine your 100 person company going to 6 people. Imagine your 1000 person company going to 20. How much faster could you move?”

I see a very real possibility of a 3-person team reaching this billion-dollar valuation.

• A CEO who distributes the thing (whilst also being proficient at building the thing).

• A CTO who builds the thing.

• A COO who manages the thing.

This triumvirate effectively covers product development, marketing, support, operations, finance and legal functions by leveraging automation, agents and APIs.

In the past, we’ve seen this trend start to take shape:

• 2008: Plenty of Fish (1 employee) generating $10M in revenue

• 2012: Instagram (13 employees) selling to Facebook for $1B

• 2014: WhatsApp (32 employees) selling to Facebook for $19B

Whilst the idea of a one-person billion-dollar company might seem like a stretch today—you’d need a hyper-viral self-serve platform—the constraints of yesterday are very different from the constraints of tomorrow.

Build once. Let it work for you. Iterate over time.

That’s the ultimate form of leverage.

We’re coming closer and closer to making this a reality.

💡 If you enjoyed this issue, share it with a friend.