Welcome back. This week has been a crazy decade in AI, with mind-blowing developments. I was particularly struck by OpenAI’s dramatic revenue projections, the surprising underperformance of leading Chinese AI models on new benchmarks, and bold predictions about AI’s transformative impact on work and society. In this newsletter, I’ll take you through the latest updates in social innovation, business strategies, media advancements, legal tech, coding trends, cybersecurity challenges, educational debates, government initiatives, robotics breakthroughs, startup dynamics, and crypto innovations.
The week’s major story is OpenAI’s newsjacking of Google’s breakthrough in its ongoing efforts to dominate the AI market. Google’s new Gemini Pro 2.5 was released, ruling the benchmark leaderboards, not just by a little. This is the model we’ve all been waiting for and know Google was capable of. Gemini 2.5 Pro leads in every benchmar category. While the Chinese released models this week too including Deepseek V3.1, rumored to be the base model for the upcoming Deepseek R2, and OpenAI beefed up with GPT4o, nobody could touch Gemini, including Elon’s Grok. However, only an hour after Google’s momentous announcement, OpenAI released its new GPT4o ImageGen tool. This is the most impressive image generator since Midjourney 1.0 and may be of all time. What is most remarkable about ImageGen is its total control over the image generation process and its nearly flawless text. Thoughts on the demise of the other image generator companies and even Adobe swirled as a result. Check out this fake WSJ article I created with it.
This week’s other AI news reminded me of stories reflecting AI’s evolving role and humanity’s desire to resist these advancements. In a blend of virtue signaling and sour grapes, Yann LeCun shared insights into how Meta had developed and then retracted a ChatGPT-like system because the public wasn’t ready to use it, contrasting sharply with public enthusiasm for ChatGPT which released a few months later. Meanwhile, DeepSeek’s upgraded model has made significant gains in reasoning and coding but is far behind Google’s Gemini 2.5 Pro, adding pressure on other established US AI giants to follow suit. Sam Altman emphasized the growing role of AI in coding, while OpenAI’s top models dramatically underperformed on the new ARC-AGI-2 benchmark, exposing significant challenges. There is also buzz around Anthropic’s potential launch of a Claude 3.7 Sonnet model with a 500K token context window, promising to transform enterprise workflows. On the creative front, ChatGPT’s image-generation upgrade now delivers more realistic visuals, and in a striking prediction, Bill Gates warned that AI might soon replace key professional roles. Finally, Microsoft’s new AI agents are expanding enterprise capabilities, while the Replit CEO’s advice to people to give up on coding as the occupation is doomed.
In business news, we heard of a significant strategic shift as IBM plans to lay off 9,000 US workers by 2025, which may pave the way for increased offshoring to India. Simultaneously, the number of million‐dollar one‐person businesses has doubled, signaling that solopreneurs are increasingly capturing significant market value. The video game industry is also feeling the pinch, shedding jobs to protect margins, and a recent IDC survey reveals that 70% of Asia/Pacific organizations expect AI to disrupt business models within the next 18 months, underscoring the rapid pace of technological change.
Innovation is front and center as Google Vids automatically adds AI-powered voiceovers to its video creation tool, streamlining multimedia content production. Meanwhile, an IAB report shows that 70% of media companies have yet to fully adopt AI, though many plan to implement it by 2026, highlighting both a challenge and an opportunity for the industry.
In the legal arena, transformative tools are emerging to reshape litigation processes. Verbit’s Legal Visor now delivers real-time deposition insights by detecting inconsistencies and enabling intelligent search during proceedings. At the same time, the DOJ is pushing for structural changes at Google by demanding the divestiture of Chrome—and possibly Android—to counter alleged monopolistic practices. Additionally, a survey reveals that 44% of legal departments already use generative AI, with further investments on the horizon.
In coding news, ScienceSoft’s research suggests that by 2026, only 20% of software engineers will work exclusively in the office, with 80% favoring remote or hybrid arrangements [ScienceSoft study]. This shift underscores the evolving dynamics of the tech workforce as companies adapt to new work models.
On the cybersecurity front, Oracle is facing a significant challenge after a hacker claimed to breach its Oracle Cloud login servers, resulting in a leak of 6 million user records.
In education, a spirited debate at Harvard examined the role of generative AI in academia. Faculty members discussed whether AI tools are a beneficial shortcut to accelerate learning or if they risk undermining fundamental skills, emphasizing the need for a balanced approach in integrating technology into academic practices.
In government initiatives, innovative approaches are emerging to enhance inmate welfare. For example, virtual reality is now being used in prisons to aid inmates in solitary confinement, potentially offering benefits for mental health and rehabilitation.
In robotics, significant updates are coming from China. Unitree Robotics has postponed the launch of its household humanoid robots to focus on safety and industrial applications. In contrast, a Chinese robot dog has been enhanced with one-touch navigation for rescue missions, improving its autonomous capabilities.
In startup news, OpenAI is gearing up for a transformative leap, projecting a tripling of its revenue to $12.7 billion in 2025. Supported by a substantial $40 billion investment from SoftBank and strategic realignments, this forecast positions OpenAI as a leading AI company.
Finally, an exciting development is underway in crypto as OpenAI CEO’s World Network is exploring a partnership with Visa. This integration aims to merge self-custody crypto wallets with Visa card features, enabling stablecoin payments across a vast merchant network and broadening fintech capabilities.
That is it. Stay tuned for more news next week. I APPRECIATE ALL THE READERS. Help spread the word. I put these newsletters together so productive humans like you can stay on top of the latest AI stories shaping our world.
If you like these weekly tech news reports, subscribe to get notified of new editions and updates. For daily updates, check out our news page. For a more in-depth analysis of the week’s news, sign up for our free weekly newsletter to the right of the daily news, or follow me on Twitter or YouTube.
