At the Google Cloud Summit in Doha, AI agents were a major focus as Qatar expands AI training and testing, raising questions over oversight and jobs.
Artificial intelligence is moving beyond generating text, images and answers. Now, developers are trying to get AI to carry out tasks on a user's behalf.
ADVERTISEMENT
ADVERTISEMENT
That shift to autonomous or "agentic AI" was a central theme at the Google Cloud Summit in Doha, where the company marked three years since opening its cloud infrastructure in Qatar.
Generative AI creates content in response to a prompt. AI agents go further, working through the steps needed to complete a task.
"Generative AI effectively is where you use AI through large language models to create net new content," Alex Rutter, Managing Director of Google Cloud's Artificial Intelligence business across Europe, the Middle East and Africa, told Euronews.
"Agentic is much more focused... on the business process workflow," he added.
From answering questions to carrying out tasks
One area businesses are looking at is repetitive work that involves checking information across different systems.
Rutter gave the example of invoice reconciliation, where information on an invoice may need to be checked against payment and supplier records before a payment is approved.
"An agent is very well suited to that sort of highly repetitive database work, where it would drive a significant level of efficiency," he said.
The more AI is allowed to do on its own, though, the more important human oversight becomes.
Rutter said Google currently advocates keeping people involved in checking the results produced by AI agents.
"We still always advocate to having at this stage a human in the loop to verify for the business the results they're looking at, but the humans are looking at the exception rather than the entire data stack," he said.
The amount of oversight needed would depend on both the task and an organisation's tolerance for risk, according to Rutter. In areas such as medicine, he said AI should support rather than replace a clinician's decision-making.
Qatar expands AI training and testing
Google Cloud used the Doha summit to mark the third anniversary of its local cloud region and announced a new national skilling programme with Qatar Digital Academy.
The programme is expected to provide more than 50,000 learning opportunities in Qatar by 2030, with training aimed at different groups including workers, executives and developers, according to Google Cloud.
Ghassan Kosta, Regional General Manager for Google Cloud in Qatar, Oman, Bahrain and Iraq, said the company was also expanding its services to support more advanced AI systems.
"We are doubling down on our services to accommodate this agentic AI era," Kosta told Euronews.
"In our data centres, to cope with how the technology and how the AI is moving, you need also the right infrastructure in place."
Another initiative announced at the summit is an innovation lab developed by Qatar's Ministry of Communications and Information Technology (MCIT) in partnership with Google Cloud.
Amna Al-Kaabi, Head of Emerging Technologies at the ministry, told Euronews the lab would allow AI projects to be tested before being put into use.
"The core value of this partnership lies in linking technology experimentation within the lab to real sector needs," Al-Kaabi said, adding that the aim was to help projects move "from proof-of-concept stage to sustainable real-world implementations."
What happens to the workforce?
That ability to hand more work to AI also brings the conversation back to jobs.
Rutter cautioned that predicting those changes remains difficult, but said he expects managing AI agents to become part of some people's jobs.
"I think what is very real is we're going to have non-people managers, we'll then have agent managers, and then we'll have people that manage people and agents together," he said.
That could mean some workers eventually managing AI agents as part of their jobs.
But Rutter acknowledged it is difficult to predict how quickly that might happen. For now, companies are still figuring out where AI agents are useful and how much human oversight they need.
