Anthropic AI helps you create your own next version
Anthropic said its artificial intelligence model Claude is actively involved in developing the next, even more intelligent version of itself.
Anthropic said its artificial intelligence model Claude is actively involved in developing the next, even more intelligent version of itself.
According to the company’s official statement, Claude already leads 26% of Anthropic’s research and development. This means that the model is capable of performing most tasks “from start to finish based on a high-level query,” while remaining under human control. The system has not yet achieved full autonomy, Tech Xplore reports .
In addition, about 90% of all the company's development takes place in "collaboration" with Claude, when the model performs large amounts of work under strict human guidance. For comparison, back in February, the share of work led by Claude was zero, and by August it had reached a quarter of all processes.
As of August, the company’s research and engineering efforts involved about 30,000 AI agents, for which Anthropic implements special monitoring measures and engages external independent evaluators to monitor their safety. At the same time, the company encourages other labs to regularly publish similar indicators so that the public has a clear understanding of how close the industry is to recursive self-improvement — the ability of AI to completely independently create its own successor.
The statement comes amid a heated debate about the safety of artificial intelligence, with Anthropic co-founder Dario Amodei and a number of other industry leaders, including Sam Altman and Elon Musk, calling for a slowdown in the pace of technology development to minimize the risks of losing control over autonomous systems.
Anthropic CEO warns of threat of entire internet being taken over by uncontrolled AI agents within a year
Anthropic CEO warns of threat of entire internet being taken over by uncontrolled AI agents within a year
"The world will be okay." Palantir co-founder urges not to panic over threat of AI destruction of humanity
"The world will be okay." Palantir co-founder urges not to panic over threat of AI destruction of humanity
China explained what they consider the main threat from artificial intelligence
China explained what they consider the main threat from artificial intelligence
Trump ignored the call of Amodei, Altman and Musk. In his opinion, the risk of losing the AI race to China is more real than the risks to human security from AI
Trump ignored the call of Amodei, Altman and Musk. In his opinion, the risk of losing the AI race to China is more real than the risks to human security from AI
OpenAI postpones IPO due to AI security risks
OpenAI postpones IPO due to AI security risks
Read the country's main IT news in our Telegram
Read the country's main IT news in our Telegram
