An AI startup, backed by Amazon and Google, has developed an AI agent capable of performing tasks on a computer, like moving a mouse cursor or entering text.
US company Anthropic has unveiled its AI model, Claude, which can now handle computing tasks such as form filling, trip planning, and website building.
Tech companies and investors see autonomous AI agents as a promising avenue for developing technology that can both surprise users and incur considerable costs to create and maintain.
Anthropic, a competitor of OpenAI’s ChatGPT, is offering developers access to its Claude 3.5 Sonnet model, acknowledging that it may have errors.
The company stated, “At this stage, it is still experimental and can be prone to errors. We are providing early access to gather feedback from developers with the expectation of rapid improvements over time.”
Anthropic has released a demo demonstrating a model that uses spreadsheet data to perform tasks, opens tabs on the screen, and completes forms on a fictional employee’s computer.
In additional demonstrations, AI agents from Anthropic planned a sunrise trip to San Francisco and created a simple website for promotion.
Microsoft recently launched a product enabling businesses to build their own autonomous AI agents, with early adopters like McKinsey using agents for customer inquiries and scheduling.
Both Anthropic and Microsoft emphasize that their products can handle mundane tasks like form completion and client inquiries, viewing them as tools to empower employees rather than threaten their jobs.
However, the OECD warns that highly skilled occupations are most at risk from AI automation, as highlighted by the organization’s 38 member countries, including the UK, Japan, Germany, the US, and Australia.
Andrew Rogoiski, from the University of Surrey’s Institute for Human-Centered AI, believes the technology industry is at a pivotal point for widespread use of autonomous agents, potentially leading to more efficient task management and job automation.
Source: www.theguardian.com
