New AI ventures, Google's rollout and strange looking chips.
It feels like AI is evolving at breakneck speed at the moment and this week has seen some big names make the headlines. To help you cut through the noise, here's what you need to know this week:
The return - Former OpenAI CTO launches new venture
Mira Murati has launched a new AI startup, Thinking Machines Lab. The company has been quietly hiring for months and has come out of stealth mode this week, to throw down a gauntlet to the likes of OpenAI and Anthropic.
Murati's new venture claims to be on a mission to make AI systems more transparent, customisable, and widely understood. Murati’s team includes ex-OpenAI, Google and Meta researchers, and the company plans to focus on multimodal AI that collaborates with users. A key focus is AI safety and human-AI collaboration: Thinking Machines Lab intends to share code, datasets, and best practices to improve AI governance and reduce misuse.
Why it matters
Transparency in AI development remains a significant challenge. Most cutting-edge AI research is concentrated within a few labs, limiting external scrutiny and public discourse. Thinking Machines Lab’s emphasis on openness could shift industry standards, pushing for more accessible, explainable, and user-driven AI models. For businesses, this could mean greater flexibility in AI adoption, fewer compliance risks, and better alignment with corporate values.
But, this comes with a caveat - Murati, while leading on the development of OpenAI's Sora video model, infamously refused to answer questions from the Wall Street Journal, about whether the model had been trained on content from sites like YouTube (scroll to 04:25 to see the interaction). Was she just throwing shade before resigning? Possibly. Or perhaps she genuinely does want to take a more ethical, responsible approach to AI development. While Thinking Machines Lab shows promise, we'll have to wait to see how committed to transparency Murati (and her investors) really are.
The rollout – Google's AI push
Google is expanding its AI capabilities across its suite of products, rolling out Gemini Deep Research to select Google Workspace tiers. This tool helps users conduct complex research by summarising industry trends, competitor insights, and customer research in minutes. Google Workspace has over 3 billion users, giving it a massive distribution advantage over standalone AI tools like ChatGPT. Google’s AI rollout extends beyond this - this week it also announced the launch of an AI co-scientist, a multi-agent AI system designed to assist researchers in hypothesis generation and scientific discovery.
Why it matters
With Google Workspace already boasting millions of users, its AI integrations enjoy an immediate competitive edge over standalone AI tools like ChatGPT. By embedding AI into the tools businesses already use daily, Google is accelerating mainstream AI adoption, making advanced research and automation accessible at scale. For businesses, this is your sign that AI is no longer an optional add-on, but is becoming a fundamental part of core software.
The ridiculous – AI-designed chips that humans don’t understand
Researchers at Princeton and the Indian Institute of Technology (IIT) have developed AI-designed wireless chips that outperform human-created designs. AI-designed chips demonstrated superior performance but often took an unconventional approach, producing designs that were highly efficient yet difficult for human engineers to decipher. Many iterations failed outright, requiring careful human oversight to identify functional designs and understand their mechanisms, while those that worked often used unconventional shapes.
Why it matters
AI-driven hardware design could revolutionise sectors reliant on high-performance chips, from telecoms to autonomous systems. But it raises an unsettling question: if engineers can’t fully understand AI-optimised components, how do we ensure reliability and security? And it's not just computer chips and hardware that are turning to AI for discoveries and design; we've seen developments in fields like protein mapping, biomedicine and drug discovery that are beginning to utilise AI.
Businesses exploring AI-generated design and discovery must weigh the trade-off between efficiency and the risks of unpredictable or unexplainable outcomes.
Final thoughts
AI’s rapid expansion brings both opportunities and challenges. Murati’s push for transparency could redefine industry norms, Google’s AI integrations signal the mainstream adoption of AI in the workplace, and AI-designed hardware raises crucial questions about control and oversight.
Businesses must stay ahead of these shifts to harness AI’s potential while mitigating its risks. If you'd like help figuring out what matters most for your business, I'd love to chat.