Anthropic, Enhanced Games, and AI: Towards a Controversial Future

Code with Claude: The Future of Coding

At a developer event organized by Anthropic in London this week, the tool Code with Claude was highlighted as a symbol of the future of coding. Participants were asked about their use of this system, and nearly half raised their hands, indicating that they had already deployed code entirely generated by Claude. What is striking is that many of them admitted to not having read the code before deploying it. This trend illustrates a growing trust in automation tools, although it raises concerns among some developers who are not convinced that this approach is the best.

With the continuous improvement of tools like Claude Code, more and more developers seem ready to delegate part of their work to artificial intelligence. Anthropic, the company behind Claude, is pushing for ever-greater automation. However, this vision is not shared by everyone, with some warning about the potential risks of such dependence on AI.

The Enhanced Games: A Controversial Competition

This Sunday, in Las Vegas, 42 athletes will gather to participate in the first Enhanced Games, a competition that has already generated significant controversy due to its nature. Unlike traditional competitions, these games allow the use of performance-enhancing drugs, with the goal of "pushing the limits of human performance." This initiative reflects a societal trend where optimizing physical performance and the quest for longevity have become priorities for many.

The event embodies a spirit of maximizing physical appearance, where participants and spectators are encouraged to become slimmer, optimize their longevity, and consider having their "best baby." By 2026, the idea of not seeking improvement could be perceived as a form of stagnation. However, this approach also sparks debates about the ethical implications and long-term consequences of using such substances to enhance human performance.

Google I/O: The Future of AI-Driven Science

During the keynote at Google I/O, Demis Hassabis, CEO of Google DeepMind, made a striking statement asserting that we are "at the foot of the singularity." This announcement was made in a context where AI is taking an increasingly important role in the scientific field. Two main directions are emerging for AI in science: on one hand, the construction of specialized systems like WeatherNext to solve specific problems, and on the other hand, the development of systems based on LLMs (language models) capable of conducting cutting-edge research projects without human intervention.

The flagship announcement of the conference was Gemini for Science, a project aimed at a future where AI-driven agents could play a central role in scientific research. While these agents may still rely on specialized systems, Google seems to be gradually moving away from them to adopt a more integrated and automated approach.

World Models: A New Understanding of Reality

Many leading AI researchers are turning to a new type of system called world models, which aims to understand the physical environment more deeply. This concept is supported by influential figures such as Fei-Fei Li from Google DeepMind and former Meta AI chief Yann LeCun. The idea behind these models is to enable AI to perceive and interact with reality in a more nuanced and precise manner.

During an exclusive roundtable, MIT Technology Review editor-in-chief Mat Honan, along with journalists Will Douglas Heaven and Grace Huckins, discussed the potential impact of these models on the future development of AI. Subscribers can now watch the full recording of this discussion, which is among the 10 things that matter in AI right now, according to our list of the most relevant topics in the field of artificial intelligence.

Must-Read

I've scoured the internet to bring you the most captivating and important stories about technology today:

• Trump postponed an order on AI due to fears of over-regulation, worrying it could become "an obstacle." (CNBC)
• OpenClaw engineers warn that a crisis of "vibe-coded slop" is imminent, with AI flooding the world with mediocre and potentially dangerous code. (WSJ)
• SpaceX canceled the launch of a new Starship prototype after discovering an issue with the ground system. (CNBC)
• Meta settled a lawsuit regarding social media addiction from a school district, which accused it of harming students. (BBC)
• Bluesky claims to have been hacked by the Kremlin to spread propaganda, fighting against Russian efforts to hijack real user accounts. (NYT)
• The largest economies in Africa are pushing for AI sovereignty, aiming to reduce their dependence on major tech companies. (Rest of World)
• Undersea cables threaten AI expansion plans in the Gulf, with conflicts jeopardizing this fragile critical infrastructure. (Wired)
• Waymo suspended its services as robotaxis continue to drive in floods. (TechCrunch)
• Microscopic silica spheres could help cool the planet, but some researchers are calling for more evidence. (The Economist)
• Spotify will now allow subscribers to create remixes with AI, a first for the platform. (Guardian)

Quote of the Day

"You have AI — true intelligence." — Steve Wozniak, co-founder of Apple, reassured graduates about the impact of AI, eliciting applause, in contrast to the boos received by former Google CEO Eric Schmidt earlier this week, according to Business Insider.

One More Thing

The future is disabled. Technologies for disability, access, and mobility are often presented as empowering solutions or heroic panaceas for social ills. However, their benefits are often temporary, uneven, or reliant on constant investment, care, and attention. Often, accessibility technology assumes levels of access that do not exist: reliable internet, smartphones, or affordable devices. Projects frequently neglect the communities they claim to serve. Yet, there is another way: embracing a total access mindset and the expertise of disabled individuals. Discover how this approach could create a more livable world for everyone.