PLUS: McDonald's AI-powered drive-thrus, a warning from Tailwind, and Gemini's arrival in your Gmail
Good morning
In a major milestone for AI reasoning, a system has successfully solved a mathematical puzzle that has stumped human experts for decades.
This achievement demonstrates AI's evolution from an information-retrieval tool to one capable of novel problem-solving. How will this ability to autonomously discover and document complex ideas reshape the future of scientific research?
In today’s Next in AI:
AI autonomously solves an unsolved math problem
McDonald's AI-powered drive-thru overhaul
A warning from Tailwind on AI's true impact
Gemini's integration into your Gmail inbox
AI's 'Eureka' Moment
Next in AI: Renowned mathematician Terrence Tao announced that an AI system has successfully solved a long-standing, unsolved mathematical puzzle. This marks a significant milestone, demonstrating AI's growing ability to tackle novel problems that lack known solutions in literature.
Decoded:
The system autonomously solved the problem even after human experts first had to reconstruct the original question from 1975, which was vaguely worded.
The solution emerged from a collaborative workflow where ChatGPT generated proofs, an AI tool named Aristotle formally verified them and fixed errors, and humans provided feedback.
Beyond just finding the answer, the AI also demonstrated the ability to rapidly write and rewrite the proof into a document approaching research-paper quality, a traditionally time-consuming task.
Why It Matters: This event shows AI evolving from an information-retrieval tool to one capable of genuine mathematical reasoning. Automating the discovery and documentation of complex ideas could significantly accelerate research and innovation across all scientific fields.
McDonald's AI Makeover
Next in AI: McDonald's is teaming up with Google Cloud to deploy a new generation of AI across its restaurants. The goal is to automate ordering, ensure order accuracy, and create a faster drive-thru experience for customers.
Decoded:
AI-powered voice bots will begin taking drive-thru orders, aiming to reduce pressure on staff and shorten wait times for customers.
New 'Accuracy Scales' will use smart technology to weigh completed orders, ensuring you get exactly what you paid for before leaving the window.
The new systems run on Google's Edge computing platform, which allows for on-site data processing to power smart devices throughout the kitchen.
Why It Matters: McDonald's massive scale makes this one of the largest real-world AI deployments in the retail space, setting a new standard for fast-food operations. This move signals how AI is moving beyond digital apps to directly change our everyday physical experiences.
The Tailwind Warning Shot
Next in AI: The popular developer tool Tailwind CSS laid off 75% of its team, sparking a debate on whether massive AI investments are overlooking the small, efficient teams that build the web's foundational tools.
Decoded:
Despite its lean team, Tailwind grew to be an industry staple, currently used by 1.5% of the web and frequently generated in AI-powered code suggestions.
The news raises questions about the return on investment after billions spent on AI in the last year, with few, if any, small teams creating new tools with the same impact as Tailwind.
Proponents argue that LLMs can't yet replicate the thoughtfully crafted work of a small team that produces a lean, high-quality framework that other businesses now build upon.
Why It Matters: This event forces a critical question on whether current AI is better at remixing what exists than creating truly new, fundamental tools. It serves as a reality check for the industry to weigh the value of large-scale AI against supporting the human-led innovation that underpins it.
Gmail's Gemini Era
Next in AI: Google is integrating Gemini AI directly into your inbox to act as a proactive assistant, sparking a broader conversation about data privacy and user control.
Decoded:
In response, Samsung is championing a hybrid AI model at CES, which processes personal data on-device to give users more control over their privacy.
The partnership's scale is massive, with Samsung planning to double the number of its mobile devices featuring Gemini-powered AI this year.
This deep integration raises major privacy questions, as the new features require cloud-based AI to scan your inbox, even if Google states it isn't using the data for model training.
Why It Matters: As AI assistants become deeply embedded in essential tools, the line between helpful automation and data intrusion gets blurrier. You now face a critical choice between leveraging powerful new features and maintaining control over your personal information.
AI Pulse
Fears grew over a potential AI bubble, with the Bank of England's governor and Alphabet's chief among those warning that tech stocks are becoming dangerously overvalued.
Researchers proposed a "surgical alignment" technique called CAST, which reduces capability loss in LLMs by selectively skipping updates to a small group of high-conflict attention heads during safety tuning.
X restricted its Grok AI image generation feature to paying subscribers, following widespread backlash and regulatory pressure over the tool's use in creating non-consensual deepfakes.