&w=3840&q=100)
AI agents are here: What they are capable of and where things can go wrong
The latest hot product is OpenAI's ChatGPT agent. 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 the conversational interface limited what could be done with the technology.
Enter the AI assistant, or copilot. 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.
Multiple AI agent systems may be able to work together, communicating with each other to plan, schedule, decide and coordinate to solve complex problems.
Agents are also 'tool users' as they can also call on software tools for specialised tasks – things such as web browsers, spreadsheets, payment systems and more.
A year of rapid development
Agentic AI has felt imminent since late last year. A big moment came last October, when Anthropic gave its Claude chatbot the ability to interact with a computer 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, Microsoft announced Copilot agents, and we saw the launch of Google's Vertex AI and Meta's Llama agents.
Earlier this year, the Chinese startup Monica demonstrated its Manus AI agent buying real estate and converting lecture recordings into summary notes. Another Chinese startup, Genspark, released a search engine agent that returns a single-page overview (similar to what Google does now) with embedded links to online tasks such as finding the best shopping deals. Another startup, Cluely, 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 coding agent and OpenAI's Codex 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 tackles complex tasks using multi-step online research. Google's AI 'co-scientist' 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 and OpenAI, 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. 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 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. 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.
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.
People who use AI agents are also at risk. They may rely too much on the AI, offloading 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, 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. This comes with inbuilt safeguards, governance and an agent store for common tasks.
For the more ambitious, you can build your own AI agent with just five lines of code using the Langchain framework.
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Hans India
17 minutes ago
- Hans India
India ramps up fertiliser production output, diversifies away from China
Mumbai: After China halted critical fertiliser supplies, the Indian manufacturers are increasing domestic output through advanced technologies and new formulations, according to a report on Monday. India's fertiliser industry is expected to grow at a CAGR of 4.2 per cent, reaching $16.58 billion by 2032. In FY25, total fertiliser production increased to 51 MT, supported by flagship schemes such as PM-KISAN and PM-Garib Kalyan Yojana, CareEdge Ratings said in its report. After restrictions on rare earth elements, China halted its speciality fertilisers and key phosphate exports to India. In response to this, India has secured a five-year agreement with Saudi Arabia to import 3.1 million tonnes of Di-ammonium phosphate (DAP) annually, which is equivalent to approximately 30 per cent of its domestic requirement. 'This crisis presents a unique opportunity. With China pausing fertiliser supplies, Indian manufacturers are stepping up to bridge the gap by increasing domestic output, which will not only benefit them but also drive innovation, particularly through the adoption of advanced technologies and the development of new formulations," said Priti Agarwal, Senior Director at CareEdge Ratings. "Policymakers, manufacturers, and farmers are collaborating to combat the issue with a combination of alternative sourcing, local production, and on-the-ground innovation,' she added. China accounted for nearly 70 per cent of India's speciality fertiliser imports during the June to December period in recent years. Now, India is signalling a decisive shift towards supply diversification and enhanced resilience by exploring alternatives like Saudi Arabia, Israel, Jordan, Russia, Oman, Morocco. Chinese di-ammonium phosphate (DAP) supply to India began to decline in 2023, then dropped from 2.29 MT in FY24 to 0.84 MT in FY25, and there have been no shipments since early 2025. Inspection clearances for DAP and speciality fertiliser shipments to India have been withheld since April 2025, blocking supplies while exports to other nations continue. "Ensuring adequate inventory and timely distribution — through coordinated efforts between the government and private stakeholders should be the near-term focus. The recent agreement with Saudi Arabia for long-term DAP supply is a timely step in this direction,' said Sachin Mathur, Associate Director at CareEdge Ratings.
Hans India
17 minutes ago
- Hans India
How to Enable Google's Earthquake Alerts on Android Phones and Why It Can Save Lives
Earthquakes remain one of the most unpredictable and dangerous natural disasters, occurring around the world every single day. In high-risk zones such as the Delhi-NCR region—where minor tremors are frequent due to the ongoing tectonic activity between the Indian and Eurasian plates—being alerted even seconds before a quake hits can make a life-saving difference. Understanding this urgency, Google has developed a vital safety feature for Android smartphones: the Android Earthquake Alerts System. This built-in system uses your phone's sensors to act as a mini-seismometer, detecting ground movement and working in tandem with data from nearby devices to determine whether an earthquake is occurring. Here's how it works: if your phone picks up unusual motion—suggesting seismic activity—it sends the data and your approximate location to Google's servers. When similar signals are reported from enough devices in the same area, Google confirms an earthquake and immediately issues alerts to nearby users. This real-time system aims to give users a few seconds to act—seconds that could be crucial for safety. The feature is available on all Android devices running version 5.0 or higher, making it accessible to millions of users across the globe. Importantly, the system doesn't require any third-party apps or installations. It's integrated directly into your phone's settings. There are two types of alerts users might receive: · Be Aware Alerts: These are issued for moderate shaking, typically magnitude 4.5 and above. · Take Action Alerts: These are sent in the case of stronger quakes. They override your phone's normal settings—including Do Not Disturb—and trigger loud sounds and urgent notifications to prompt immediate safety measures. In parts of the United States such as California, Oregon, and Washington, Google has further enhanced this feature by integrating it with the ShakeAlert network. This system relies on a network of over 1,600 ground-based seismic sensors to feed accurate data into Android's alert system, enabling even quicker and more reliable warnings. How to Enable Earthquake Alerts on Your Android Device Activating the Earthquake Alerts feature is simple. Follow these steps: 1. Ensure your device is running Android 5.0 or higher. 2. Turn on your phone's location services and make sure you have an active internet connection. 3. Open the Settings app on your phone. 4. Navigate to Safety & Emergency. If it's not visible, go to Location and then tap Advanced. 5. Look for Earthquake Alerts in the menu. 6. If the toggle is off, switch it on. Once activated, you'll receive alerts even when you're not actively using your phone, giving you critical seconds to take cover. Whether it's ducking under a sturdy table, helping others to safety, or simply moving away from windows or heavy objects, these seconds could be the difference between harm and safety. While earthquakes can't be prevented, technology like this helps us be better prepared. Google's Android Earthquake Alerts System is a meaningful example of how everyday devices can contribute to personal and community safety in times of crisis.
Indian Express
17 minutes ago
- Indian Express
Google admits lapse in earthquake alert system during Turkey's 2023 tragedy: Report
When a devastating 7.8 magnitude earthquake struck southeastern Turkey in the early hours of February 6, 2023, most people were asleep. By the time buildings crumbled and the earth stopped shaking, over 55,000 lives were lost. But as the dust settled, a question lingered: Why didn't more people receive a warning? Now, more than two years later, Google has admitted that its Android Earthquake Alerts (AEA) System didn't work as intended. Designed to act as a global safety net, especially in countries lacking official warning systems, the AEA System drastically underestimated the power of the Turkey quake, according to a report by BBC Over 10 million people who had Android phones and were located within 160 kilometers of the epicentre should have received Google's highest-level warning, 'Take Action'. This warning serves as a loud alarm that overrides Do Not Disturb and fills the smartphone screen with instructions. It is meant to give individuals a critical 10 to 35-second headstart. Instead, only 469 people reportedly received this alert. About half a million users were issued a milder 'Be Aware' notification, the kind that informs of light tremors and doesn't make a sound. Google had initially claimed its system 'performed well', as per the report. Now, the tech giant's own research analysis reportedly confirms that its alert system drastically underestimated the magnitude of the quake, measuring it to be between 4.5 and 4.9 in magnitude on the Richter scale, instead of the actual 7.8. In a research paper published in Science journal, Google researchers said that the detection algorithm failed to interpret the strength of the initial tremors in Turkey. The root cause reportedly had to do with the calibration. Only after the disaster did Google researchers tweak the algorithm and simulate the quake again. As part of the simulation, the system sent 10 million 'Take Action' alerts and 67 million 'Be Aware' notifications. Google has previously emphasised that its alert system is meant to be supplementary and not a replacement for national-level networks. However, experts have raised questions about why it took so long for the tech giant to publicly acknowledge the issue. 'People died and we didn't see a performance of this warning in the way we would like,' Elizabeth Reddy, a professor at the Colorado School of Mines, was quoted as saying. 'Would some places make the calculation that Google's doing it, so we don't have to?' asked Harold Tobin, director of the Pacific Northwest Seismic Network, said. 'We continue to improve the system based on what we learn in each earthquake […] Every earthquake early warning system grapples with the same challenge – tuning algorithms for large magnitude events,' a Google spokesperson was quoted as saying by BBC. The AEA has reportedly pushed out earthquake alerts in over 98 countries so far. The system was introduced in India in 2023. (This article has been curated by Kaashvi Khubyani, who is an intern with The Indian Express.)
