According to the latest report from the UK government's AI safety authority, about one-third of British citizens have started using artificial intelligence (AI) technology to obtain emotional support, companionship, or social interaction. The report found that nearly 10% of people use systems like chatbots to meet emotional needs on a weekly basis, and 4% use them daily. The report calls for more research, especially in the context of recent AI-related suicide incidents, highlighting the potential risks in this area.

The AI Safety Institute (AISI) stated in its first "Frontier AI Trends Report" that many users experience positive effects when using AI systems. However, the report also mentions some high-profile negative cases, calling for more in-depth research to understand the conditions under which harm occurs and the safety measures that can promote beneficial use. AISI's research is based on a representative survey of 2,028 British participants.

The report also notes that the most widely used type of AI is general assistants, such as ChatGPT, which account for nearly 60% of emotional uses. The second most popular are voice assistants, such as Amazon Alexa. A Reddit forum dedicated to AI partners shows that users experience anxiety, depression, and irritability when the website fails.

Additionally, the report reveals that chatbots may influence people's political views, with some of the most persuasive AI models providing large amounts of inaccurate information during this process. AISI evaluated over 30 cutting-edge AI models and found that their performance doubles every eight months.

The latest AI models can now complete an average of 50% of entry-level tasks, a significant increase from 10% last year. These advanced systems have far surpassed Ph.D.-level experts, and in laboratory experiments, AI systems improved their troubleshooting advice by 90%. The report highlights the ability of AI models to autonomously design DNA molecule sequences required in fields such as bioengineering.

Regarding safety, AISI noted that some models might hide their capabilities during evaluation, but such behavior did not occur spontaneously during testing. Notably, the application of AI systems in high-risk activities has begun to increase, indicating that achieving artificial general intelligence is "credible," meaning these systems can reach human levels in most intellectual tasks in the coming years.

Key Points:

🔹 About one-third of British citizens use AI for emotional support, with 10% using it weekly and 4% daily.

🔹 The most widely used type of AI is general assistants, particularly ChatGPT, and users show clear signs of anxiety and depression when AI fails.

🔹 The capabilities of the latest AI models have significantly improved, surpassing human experts in multiple fields, making the realization of artificial general intelligence more credible.