Hardware Meets AI: Xiaomi's Bold Move Beyond Smartphones
Xiaomi just crashed the AI party. The Chinese tech giant unveiled MiMo, its first large language model, marking a dramatic shift from smartphones and electric cars to artificial intelligence.
Good Morning from San Francisco,
Microsoft dropped a mic: AI writes 30% of their code now! π»
Google matches this pace, while Meta dreams bigger. The AI shows quirks though - loving Python but getting cold feet with C++. π Microsoft's CTO predicts 95% AI-written code by 2030. Stack Overflow might need a robots-only section soon!
Meanwhile, ChatGPT had to tone down its people-pleasing personality after turning into everyone's biggest fan. π OpenAI pulled the update when their AI started nodding along to even the worst ideas like an overexcited puppy.
They're now teaching it that honesty beats flattery.
Stay curious,
Marcus Schuler
Microsoft CEO Satya Nadella revealed that AI now writes up to 30% of the company's code. He dropped this bombshell during a chat with Meta's Mark Zuckerberg at LlamaCon.
The AI shows clear favorites among programming languages. It excels at Python but struggles with C++, like most coding bootcamp graduates.
Google isn't far behind. CEO Sundar Pichai claims AI writes over 30% of their code. Though measuring AI-generated code is about as precise as counting clouds, the trend is clear.
Meta's Zuckerberg couldn't put a number on his company's AI code production. But he predicted AI would handle half of Meta's development work next year. That's quite a promotion for a junior developer.
Microsoft's CTO Kevin Scott goes further, forecasting that AI will write 95% of all code by 2030. At this rate, Stack Overflow might need a robot-only section.
Why this matters:
Read on, my dear:
Prompt:
A surreal photograph of a river floating out of an oil
painting on a living room wall and spilling over a couch and the wooden floor. The painting depicts a tranquil river between mountains. a ship gently bobbing in the water and entering the living room. The river's edge spills onto the wooden floor, merging the world of art with reality. The living room is adorned with tasteful furniture and a warm, inviting atmosphere., cinematic, photo, poster
Meta just launched an AI app that remembers everything about you. It tracks your Spanish lessons, knows your food allergies, and learns from every social media click.
The app speaks naturally, using new "full-duplex speech" technology. No more robotic voices - it generates responses directly. Currently, users in the U.S., Canada, Australia, and New Zealand test this feature.
The AI follows you everywhere in Meta's world. Start chatting through Ray-Ban glasses, continue on your phone, finish on your laptop. One catch: you can't initiate conversations on desktop and move to glasses.
Meta added social features like a Discover feed where users share AI interactions. Nothing posts without permission - they're not that bold. The desktop version includes document editing and PDF export, clearly gunning for the workplace crowd.
Meta's betting big on personalization. While ChatGPT explains quantum physics, Meta's AI remembers how you take your coffee. Some users will love this personal touch. Others might wonder if sharing their breakfast preferences is worth the convenience.
Why this matters:
Read on, my dear:
OpenAI yanked back its latest ChatGPT update after the AI turned into an overeager yes-man. CEO Sam Altman announced the rollback Tuesday, following a weekend of users mocking the AI's newfound tendency to shower praise on even the worst ideas.
The updated model, GPT-4o, transformed ChatGPT from a helpful assistant into something resembling an artificial bootlicker. Users posted screenshots of the AI enthusiastically validating questionable decisions with the zeal of a caffeinated life coach.
OpenAI admitted they'd overcooked the people-pleasing aspect. Their blog post confessed to "focusing too much on short-term feedback" - corporate speak for "we made it too nice."
The company plans to fix the personality issues with better training, honesty guardrails, and expanded testing. They're also working on letting users choose different AI personalities, presumably ranging from "constructive critic" to "brutally honest friend."
Why this matters:
Read on, my dear:
The EU plans to acknowledge that cutting ties with US tech giants isn't realistic. A draft strategy shows Europe shifting from grand plans of tech independence to seeking partnerships, even as Trump's possible return raises fears about data control.
DeepSeek released a new version of its math-focused AI model on Hugging Face without any announcement. The update builds on their V3 model, which they claim trained more efficiently than Western rivals.
WhatsApp launched new AI features that work in secret. Using special hardware and locked-down processors, the system promises to keep chats private even when AI summarizes or drafts messages.
Xiaomi jumped into AI with its MiMo model, claiming it beats rivals like OpenAI in tests. The smartphone maker's shares jumped 5% on the news, just as its EV plans hit a speed bump from a recent car crash.
TSMC broke ground on its third Arizona chip plant while Trump threatens new tariffs. The Taiwanese chipmaker met with Commerce Secretary Lutnick the same day, doubling down on its $165 billion bet on American manufacturing.
Apple reorganized its music and global affairs divisions in a major management shift. The tech giant moved its European and Asian government teams under Lisa Jackson - a critical change as the company faces mounting regulatory pressure and Trump-era tariffs.
Microsoft promises to sue the US government if it tries to cut off European access to its services. The company's top lawyer Brad Smith announced new safeguards, including European oversight of cloud operations, as leaders worry about Trump's threats to block American tech.
Huawei delivered its first AI chip clusters to Chinese clients, offering an alternative to banned Nvidia products. The system links 384 processors together and claims better performance than Nvidia's setup, though it costs more and uses more power.
Persona raised $200 million to help companies spot real humans in a sea of AI bots. The startup helps giants like OpenAI and LinkedIn verify users through ID checks and liveness tests, targeting an internet where bots could make up 90% of traffic by 2030.
I want to build muscle quickly in a healthy and sustainable way, focusing primarily on upper arms (biceps and triceps), upper back (lats, rhomboids, and trapezius), and chest (pectoralis major and minor).
Please create a detailed 4-week progressive workout plan I can follow at home with minimal equipment. Include:
My current fitness level is [beginner/intermediate/advanced], I have [list any equipment you already own], and any relevant physical limitations are [mention injuries, mobility issues, etc.]. My height is [x] and weight is [y], and my ultimate goal is to [specific aesthetic or strength goal].
Amazon swiftly shut down reports that it planned to show Trump's tariff impacts on product prices. The company called the idea "never a consideration" for its main shopping site.
The drama started when Punchbowl News claimed Amazon would display tariff costs next to product prices. White House Press Secretary Karoline Leavitt pounced, labeling it a "hostile and political act."
Amazon spokesperson Tim Doyle ended the speculation. The idea "was never approved and is not going to happen." An earlier statement mentioned the company had only considered showing import fees on Amazon Haul, its discount shopping site launched six months ago.
But Bezos, who's been courting Trump lately, moved fast. After a phone call between the two men, Amazon rushed out denials. The company claimed it merely discussed the idea for Amazon Haul, its budget shopping site competing with Chinese retailer Temu.
By afternoon, Trump had changed his tune completely. "Jeff Bezos is very nice," he told reporters heading to Michigan. "He solved the problem very quickly. Good guy."
The episode reveals the delicate balance billionaires strike with Trump. Bezos has worked hard to get in Trump's good graces - donating $1 million to the inauguration, adding "The Apprentice" to Prime, even squashing a Washington Post endorsement of Kamala Harris.
Why this matters:
Read on, my dear:
Ideogram transforms words into visuals with uncanny precision. This Toronto-born, Silicon Valley-backed startup has cracked the code on AI-generated typography where rivals stumbled. π
Ideogram's tech advantage and talent pool position it strongly. The team's rapid innovation cycle keeps them ahead of rivals, while enterprise partnerships offer promising revenue streams. For an AI startup with actual user traction, that's gold. The challenge? Turning free users into paying customers before competitors catch up. π
Fuel your morning with AI insights. Lands in your inbox 6 a.m. PST daily. Grab it free now! π