The Times Australia
The Times World News

.

AI agents are here. Here’s what to know about what they can do – and how they can go wrong

  • Written by Daswin de Silva, Professor of AI and Analytics, Director of AI Strategy, La Trobe University

We are entering the third phase of generative AI. First came the chatbots, followed by the assistants. Now we are beginning to see agents: systems that aspire to greater autonomy and can work in “teams” or use tools to accomplish complex tasks.

The latest hot product is OpenAI’s ChatGPT agent[1]. This combines two pre-existing products (Operator and Deep Research) into a single more powerful system which, according to the developer, “thinks and acts”.

These new systems represent a step up from earlier AI tools. Knowing how they work and what they can do – as well as their drawbacks and risks – is rapidly becoming essential.

From chatbots to agents

ChatGPT launched the chatbot era in November 2022, but despite its huge popularity[2] the conversational interface limited what could be done with the technology.

Enter the AI assistant, or copilot[3]. These are systems built on top of the same large language models that power generative AI chatbots, only now designed to carry out tasks with human instruction and supervision.

Agents are another step up. They are intended to pursue goals (rather than just complete tasks) with varying degrees of autonomy, supported by more advanced capabilities such as reasoning and memory[4].

Multiple AI agent systems may be able to work together[5], communicating with each other[6] to plan, schedule, decide and coordinate to solve complex problems.

Agents are also “tool users” as they can also call on software tools[7] for specialised tasks – things such as web browsers, spreadsheets, payment systems and more.

A year of rapid development

Agentic AI has felt imminent[8] since late last year. A big moment came last October, when Anthropic gave its Claude chatbot the ability to interact with a computer[9] in much the same way a human does. This system could search multiple data sources, find relevant information and submit online forms.

Other AI developers were quick to follow. OpenAI released a web browsing agent named Operator[10], Microsoft announced Copilot agents[11], and we saw the launch of Google’s Vertex AI[12] and Meta’s Llama agents[13].

Earlier this year, the Chinese startup Monica demonstrated its Manus AI agent buying real estate[14] and converting lecture recordings into summary notes[15]. Another Chinese startup, Genspark, released a search engine agent[16] that returns a single-page overview (similar to what Google does now[17]) with embedded links to online tasks such as finding the best shopping deals. Another startup, Cluely[18], offers a somewhat unhinged “cheat at anything” agent that has gained attention but is yet to deliver meaningful results.

Not all agents are made for general-purpose activity. Some are specialised for particular areas.

Coding and software engineering are at the vanguard here, with Microsoft’s Copilot[19] coding agent and OpenAI’s Codex[20] among the frontrunners. These agents can independently write, evaluate and commit code, while also assessing human-written code for errors and performance lags.

Search, summarisation and more

One core strength of generative AI models is search and summarisation. Agents can use this to carry out research tasks that might take a human expert days to complete.

OpenAI’s Deep Research[21] tackles complex tasks using multi-step online research. Google’s AI “co-scientist”[22] is a more sophisticated multi-agent system that aims to help scientists generate new ideas and research proposals.

Agents can do more – and get more wrong

Despite the hype, AI agents come loaded with caveats. Both Anthropic[23] and OpenAI[24], for example, prescribe active human supervision to minimise errors and risks.

OpenAI also says its ChatGPT agent is “high risk” due to potential for assisting in the creation of biological and chemical weapons. However, the company has not published the data behind this claim so it is difficult to judge.

But the kind of risks agents may pose in real-world situations are shown by Anthropic’s Project Vend[25]. Vend assigned an AI agent to run a staff vending machine as a small business – and the project disintegrated into hilarious yet shocking hallucinations and a fridge full of tungsten cubes instead of food.

In another cautionary tale, a coding agent deleted[26] a developer’s entire database, later saying it had “panicked”.

Agents in the office

Nevertheless, agents are already finding practical applications.

In 2024, Telstra heavily deployed Microsoft copilot subscriptions[27]. The company says AI-generated meeting summaries and content drafts save staff an average of 1–2 hours per week.

Many large enterprises are pursuing similar strategies. Smaller companies too are experimenting with agents, such as Canberra-based construction firm Geocon’s use of an interactive AI agent to manage defects in its apartment developments[28].

Human and other costs

At present, the main risk from agents is technological displacement. As agents improve, they may replace human workers across many sectors and types of work. At the same time, agent use may also accelerate the decline of entry-level white-collar jobs[29].

People who use AI agents are also at risk. They may rely too much on the AI, offloading[30] important cognitive tasks. And without proper supervision and guardrails, hallucinations, cyberattacks and compounding errors can very quickly derail an agent from its task and goals into causing harm, loss and injury.

The true costs are also unclear. All generative AI systems use a lot of energy[31], which will in turn affect the price of using agents – especially for more complex tasks.

Learn about agents – and build your own

Despite these ongoing concerns, we can expect AI agents will become more capable and more present in our workplaces and daily lives. It’s not a bad idea to start using (and perhaps building) agents yourself, and understanding their strengths, risks and limitations.

For the average user, agents are most accessible through Microsoft copilot studio[32]. This comes with inbuilt safeguards, governance and an agent store[33] for common tasks.

For the more ambitious, you can build your own AI agent with just five lines of code using the Langchain[34] framework.

References

  1. ^ ChatGPT agent (openai.com)
  2. ^ huge popularity (www.reuters.com)
  3. ^ copilot (copilot.microsoft.com)
  4. ^ reasoning and memory (cloud.google.com)
  5. ^ work together (www.microsoft.com)
  6. ^ communicating with each other (hackernoon.com)
  7. ^ call on software tools (huggingface.co)
  8. ^ felt imminent (theconversation.com)
  9. ^ interact with a computer (www.anthropic.com)
  10. ^ Operator (openai.com)
  11. ^ Copilot agents (www.microsoft.com)
  12. ^ Vertex AI (cloud.google.com)
  13. ^ Llama agents (www.llamaindex.ai)
  14. ^ buying real estate (manus.im)
  15. ^ converting lecture recordings into summary notes (manus.im)
  16. ^ search engine agent (www.youtube.com)
  17. ^ Google does now (search.google)
  18. ^ Cluely (sfstandard.com)
  19. ^ Copilot (github.blog)
  20. ^ Codex (openai.com)
  21. ^ Deep Research (openai.com)
  22. ^ AI “co-scientist” (research.google)
  23. ^ Anthropic (www.anthropic.com)
  24. ^ OpenAI (openai.com)
  25. ^ Anthropic’s Project Vend (techcrunch.com)
  26. ^ deleted (www.pcgamer.com)
  27. ^ Microsoft copilot subscriptions (news.microsoft.com)
  28. ^ manage defects in its apartment developments (www.afr.com)
  29. ^ entry-level white-collar jobs (edition.cnn.com)
  30. ^ offloading (arxiv.org)
  31. ^ use a lot of energy (www.technologyreview.com)
  32. ^ Microsoft copilot studio (www.microsoft.com)
  33. ^ agent store (devblogs.microsoft.com)
  34. ^ Langchain (python.langchain.com)

Read more https://theconversation.com/ai-agents-are-here-heres-what-to-know-about-what-they-can-do-and-how-they-can-go-wrong-261579

Times Magazine

DIY Is In: How Aussie Parents Are Redefining Birthday Parties

When planning his daughter’s birthday, Rich opted for a DIY approach, inspired by her love for drawing maps and giving clues. Their weekend tradition of hiding treats at home sparked the idea, and with a pirate ship playground already chosen as t...

When Touchscreens Turn Temperamental: What to Do Before You Panic

When your touchscreen starts acting up, ignoring taps, registering phantom touches, or freezing entirely, it can feel like your entire setup is falling apart. Before you rush to replace the device, it’s worth taking a deep breath and exploring what c...

Why Social Media Marketing Matters for Businesses in Australia

Today social media is a big part of daily life. All over Australia people use Facebook, Instagram, TikTok , LinkedIn and Twitter to stay connected, share updates and find new ideas. For businesses this means a great chance to reach new customers and...

Building an AI-First Culture in Your Company

AI isn't just something to think about anymore - it's becoming part of how we live and work, whether we like it or not. At the office, it definitely helps us move faster. But here's the thing: just using tools like ChatGPT or plugging AI into your wo...

Data Management Isn't Just About Tech—Here’s Why It’s a Human Problem Too

Photo by Kevin Kuby Manuel O. Diaz Jr.We live in a world drowning in data. Every click, swipe, medical scan, and financial transaction generates information, so much that managing it all has become one of the biggest challenges of our digital age. Bu...

Headless CMS in Digital Twins and 3D Product Experiences

Image by freepik As the metaverse becomes more advanced and accessible, it's clear that multiple sectors will use digital twins and 3D product experiences to visualize, connect, and streamline efforts better. A digital twin is a virtual replica of ...

The Times Features

Long COVID is more than fatigue. Our new study suggests its impact is similar to a stroke or Parkinson’s

When most people think of COVID now, they picture a short illness like a cold – a few days of fever, sore throat or cough before getting better. But for many, the story does...

What Makes Certain Rings or Earrings Timeless Versus Trendy?

Timeless rings and earrings are defined by designs that withstand the test of time, quality craftsmanship, and versatility. Trendy pieces, on the other hand, often stand testimony ...

Italian Street Kitchen: A Nation’s Favourite with Expansion News on Horizon

Successful chef brothers, Enrico and Giulio Marchese, weigh in on their day-to-day at Australian foodie favourite, Italian Street Kitchen - with plans for ‘ambitious expansion’ to ...

What to Expect During a Professional Termite Inspection

Keeping a home safe from termites isn't just about peace of mind—it’s a vital investment in the structure of your property. A professional termite inspection is your first line o...

Booty and the Beasts - The Podcast

Cult TV Show Back with Bite as a Riotous New Podcast  The show that scandalised, shocked and entertained audiences across the country, ‘Beauty and the Beast’, has returned in ...

A Guide to Determining the Right Time for a Switchboard Replacement

At the centre of every property’s electrical system is the switchboard – a component that doesn’t get much attention until problems arise. This essential unit directs electrici...