{"id":62396,"date":"2026-06-04T16:29:07","date_gmt":"2026-06-04T16:29:07","guid":{"rendered":"https:\/\/www.europesays.com\/ai\/62396\/"},"modified":"2026-06-04T16:29:07","modified_gmt":"2026-06-04T16:29:07","slug":"what-are-a-i-agents-actually-doing","status":"publish","type":"post","link":"https:\/\/www.europesays.com\/ai\/62396\/","title":{"rendered":"What Are A.I. Agents Actually Doing?"},"content":{"rendered":"<p class=\"css-ac37hb evys1bk0\">When OpenAI <a class=\"css-yywogo\" href=\"https:\/\/www.nytimes.com\/2022\/12\/10\/technology\/ai-chat-bot-chatgpt.html\" title=\"\" rel=\"nofollow noopener\" target=\"_blank\">unveiled ChatGPT at the end of 2022<\/a>, it kicked off the chatbot boom. Then last year, new systems from OpenAI and Anthropic incited a new technological push with so-called <a class=\"css-yywogo\" href=\"https:\/\/www.nytimes.com\/2026\/03\/19\/technology\/ai-agents-uses.html\" title=\"\" rel=\"nofollow noopener\" target=\"_blank\">A.I. agents<\/a> that can perform tasks like personal digital helpers.<\/p>\n<p class=\"css-ac37hb evys1bk0\">Now, a San Francisco start-up called Arena, which tracks hundreds of thousands of artificial intelligence users, is trying to take some of the mystery out of what exactly those digital tasks are.<\/p>\n<p class=\"css-ac37hb evys1bk0\">The company\u2019s service, Agent Mode, showed that over the past few weeks, people used agents for code-writing tasks about 17 percent of the time. Roughly 10 percent of the time, the company said, people used agents to do research.<\/p>\n<p class=\"css-ac37hb evys1bk0\">Research agents were closely followed by agents that create images, generate documents like graphs and spreadsheets or brainstorm ideas. About 5 percent of the time, the users applied agents for creative writing or tutoring and education. Other areas included code debugging, which is related to building software, and chatting.<\/p>\n<p class=\"css-ac37hb evys1bk0\">Systems from OpenAI, Anthropic and other companies can generate, test and edit computer code, letting experienced programmers <a class=\"css-yywogo\" href=\"https:\/\/www.nytimes.com\/2026\/02\/20\/technology\/ai-coding-software-jobs.html\" title=\"\" rel=\"nofollow noopener\" target=\"_blank\">automate many tasks<\/a> they once performed on their own. Agents can also <a class=\"css-yywogo\" href=\"https:\/\/www.nytimes.com\/2025\/02\/02\/technology\/openai-deep-research-tool.html\" title=\"\" rel=\"nofollow noopener\" target=\"_blank\">spend minutes or even days researching<\/a> specific topics via the wider internet, including finance, health care, the law and practically anything else.<\/p>\n<p class=\"css-ac37hb evys1bk0\">Some of these tasks overlap with what a chatbot can do. But the main difference with an agent is that it can use other software apps on behalf of users, including spreadsheets, calendars and email programs.<\/p>\n<p class=\"css-ac37hb evys1bk0\">\u201cAn agent can access the internet, search the web, create files and even access other A.I. models to complete its work,\u201d said Arena\u2019s chief executive, Anastasios Angelopoulos, a co-founder of the start-up.<\/p>\n<p class=\"css-ac37hb evys1bk0\">In Silicon Valley, some people treat these bots almost as employees they can delegate work to at any time of day. Many A.I. researchers, tech executives and pundits believe that agents could soon replace white-collar office workers.<\/p>\n<p class=\"css-ac37hb evys1bk0\">In February, Block, the financial technology company that owns Square, Cash App and Tidal, <a class=\"css-yywogo\" href=\"https:\/\/www.nytimes.com\/2026\/02\/26\/technology\/block-square-job-cuts-ai.html\" title=\"\" rel=\"nofollow noopener\" target=\"_blank\">said<\/a> it was cutting 40 percent of its work force as it anticipated the rise of this kind of technology. This was perhaps the most striking example of a company\u2019s eliminating employees because of what A.I. may soon do.<\/p>\n<p class=\"css-ac37hb evys1bk0\">The rub is that this digital employee can handle only some tasks \u2014 and sometimes, it is less than reliable. Like chatbots, A.I. agents can <a class=\"css-yywogo\" href=\"https:\/\/www.nytimes.com\/2023\/11\/06\/technology\/chatbots-hallucination-rates.html\" title=\"\" rel=\"nofollow noopener\" target=\"_blank\">make mistakes<\/a> and exhibit completely unexpected behavior.<\/p>\n<p class=\"css-ac37hb evys1bk0\">These mistakes can get particularly dicey when people use agents <a class=\"css-yywogo\" href=\"https:\/\/www.nytimes.com\/2026\/03\/19\/technology\/ai-agents-uses.html\" title=\"\" rel=\"nofollow noopener\" target=\"_blank\">to send emails, texts and other instant messages<\/a>. For that reason, Arena does not allow the people it tracks to connect their agents to email programs and messaging apps. (The company is selling its data and analysis of that data.)<\/p>\n<p class=\"css-ac37hb evys1bk0\">The company also prevents people from using agents outside a digital \u201csandbox,\u201d which prevents agents from doing serious harm on people\u2019s computers. If let outside a sandbox, agents can accidentally delete files and software apps.<\/p>\n<p class=\"css-ac37hb evys1bk0\">But the company\u2019s service gives an indication of how often agents get things wrong. About 8 percent of the time, agents said they had completed a task when they hadn\u2019t, Arena said. Because many tasks build on one another, the company added, this kind of agent \u201cbluffing\u201d or \u201cblustering\u201d can compound and create greater errors.<\/p>\n<p class=\"css-ac37hb evys1bk0\">\u201cThe models will just say, \u2018Yeah, I did this.\u2019 But they lied, and they didn\u2019t do it,\u201d Mr. Angelopoulos said. \u201cThey might say they created a file, and then it\u2019s not there.\u201d<\/p>\n<p class=\"css-ac37hb evys1bk0\">Arena also compares the technologies offered by OpenAI, Anthropic and other companies. The most effective agents are driven by OpenAI\u2019s GPT-5.5 High technology, according to Arena\u2019s data.<\/p>\n<p class=\"css-ac37hb evys1bk0\">The next most effective technology was Anthropic\u2019s Claude Opus 4.7 Thinking. These technologies, Arena said, were significantly more effective than those from Google, the leading Chinese companies and Elon Musk\u2019s xAI.<\/p>\n","protected":false},"excerpt":{"rendered":"When OpenAI unveiled ChatGPT at the end of 2022, it kicked off the chatbot boom. Then last year,&hellip;\n","protected":false},"author":2,"featured_media":62397,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[7],"tags":[1556,35679,25,1555,157,1588,1585],"class_list":["post-62396","post","type-post","status-publish","format-standard","has-post-thumbnail","category-openai","tag-anthropic-ai-llc","tag-arena","tag-artificial-intelligence","tag-computers-and-the-internet","tag-openai","tag-openai-labs","tag-start-ups"],"_links":{"self":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/62396","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/comments?post=62396"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/62396\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media\/62397"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media?parent=62396"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/categories?post=62396"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/tags?post=62396"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}