Developing a beneficial and trustworthy AI – eKathimerini.com

EmTech Europe 2026, the flagship technology conference of MIT Technology Review, was hosted in Athens for the third time earlier this week, bringing together leading voices in artificial intelligence and emerging technologies.
Organized in partnership with Kathimerini, the event heard from researchers, industry executives and innovators who focused on the evolution of AI, business and industrial transformation, and strengthening infrastructure against increasingly sophisticated cyber threats. Talks also centered on ethics and trust in the age of generative AI.
Digital Governance Minister Dimitris Papastergiou told the conference that Greece is aiming to harness artificial intelligence to modernize governance and strengthen democratic institutions. He cited data showing Greece ranks first in Europe in the use of general AI applications among people aged 16-24 and eighth overall. The challenge, he said, is turning that momentum into tangible results.
Papastergiou stressed the need to use AI to support democracy, warning that “democracy is under pressure and it is critical to help it.” He pointed to a program to simplify legislation and make it more accessible to citizens.
On infrastructure, the minister said Greece has presented a national data center strategy and plans to release thousands of public datasets by the end of the month, supported by the European Union’s Recovery Fund. He cautioned, however, that expansion must not strain the national energy grid, noting that data centers currently account for about 0.1% of energy use.
In another panel, Shakir Mohamed, research director at DeepMind, the British-American artificial intelligence research laboratory, outlined key pathways for socially beneficial and trustworthy artificial intelligence.
Speaking in a session moderated by Charlotte Jee, senior reporter at MIT Technology Review, Mohamed stressed that the future of AI is not fixed but shaped by societal choices and levels of public trust. He opened with a scenario exercise inviting audiences to consider four possible futures for AI based on two axes: technological development by 2040 and societal trust by mid-century. The exercise, he said, underscores that no single trajectory is inevitable, while pointing to the need to build “high-trust AI.”
Mohamed outlined three key application areas. In education, he described a vision of “one teacher for every student and one assistant for every educator,” enabled by AI systems grounded in established pedagogical principles such as active learning, cognitive load management and personalization.
On climate and environmental management, he highlighted advances in AI-driven weather forecasting, noting that medium-term predictions are critical for energy systems, renewable energy source integration and early warning mechanisms. The third area focused on collaborative AI, with systems such as the Aeneas network enabling historians to connect ancient texts and retrieve parallels within seconds.
Jurgi Camblong said future progress depends on ‘collective intelligence’ and data sharing, adding that ‘today’s patients help tomorrow’s patients’
He added that trust in AI mirrors broader social inequalities and must be built through “small, repeated, reliable acts.”
The growing role of artificial intelligence in transforming healthcare toward prediction, prevention and personalized treatment was the focus of another panel, moderated by Tom Ellis, the editor in chief of Kathimerini’s English Edition. It featured Jurgi Camblong, CEO and founder of Sophia Genetics, and Kira Radinsky, CTO and co-founder of Diagnostic Robotics.
Camblong highlighted the rise of personalized (precision) medicine, noting nearly 20 million new cancer cases globally in 2020 and the rapid growth of health data, particularly from genomics. He stressed that applying AI effectively requires deep biological understanding and better integration of fragmented health data. Drawing on Sophia Genetics’ work with 2.4 million patient datasets, he said future progress depends on “collective intelligence” and data sharing, adding that “today’s patients help tomorrow’s patients.”
Radinsky focused on predictive models that combine diverse data sources to anticipate health risks. She cited the ability to forecast outbreaks such as cholera by linking environmental and social factors, enabling earlier intervention. She also pointed to AI-driven preventive care systems that guide patients toward timely action.
The discussion also addressed challenges around trust and transparency. Radinsky warned of potential biases in AI models, while Camblong underscored the need for confidence in algorithmic outputs despite limits in understanding how they function.
In another panel, cybercrime expert Robert McArdle warned that artificial intelligence is helping criminals massively expand and improve their activity in ransomware and phishing attacks. He said that AI helps criminals to fine-tune the focus of their attacks and create text that appears to be entirely authentic.
McArdle, director of FTR cybercrime research at Trend Micro, also warned that deepfake visual material currently available represents the lowest quality that will ever be seen and will improve drastically in the future. He added that Agentic AI is expected to revolutionize cybercrime, which will become ever more adaptive, resilient and profitable, and authorities will be forced to deal with the agents, too, as well as the human beings behind them.
Keri Pearlson, a cybersecurity expert at MIT Sloan, said company data breaches continue to increase and traditional defenses are no longer adequate to deal with the issue. She stressed that humans remain the usual gateway for cyberattacks, by clicking on phishing emails or failing to change their passwords. Another highlight of the conference was Turing Award-winner Joseph Sifakis, who described AI as “a milestone for humanity,” while cautioning against blindly following the vision of major tech companies. “The problem is not AI, but our ability to use it,” the respected Greek-French computing scientist told the conference.
In a discussion moderated by Kathimerini’s Iliana Magra, Sifakis said that despite the strides made in developing the technology, especially in recent years, “AI is still in its infancy.” Focusing on autonomous systems, he said the goal is to replace human functions with machines capable of adapting to dynamic environments. Yet this vision faces unresolved technical challenges, including the need for extremely high reliability. “We cannot trust black boxes, and we cannot guarantee their safety,” he said.
On AI’s societal impact, Sifakis cautioned about the risk of unemployment and the weakening of state power, as governments become increasingly dependent on private tech companies. “The benefits of AI will be proportional to skills,” he said.
Enter your information below to receive our weekly newsletters with the latest insights, opinion pieces and current events straight to your inbox.

PROPERTY OF: NEES KATHIMERINES EKDOSEIS SINGLE MEMBER S.A. © 2014 – 2026
Powered by

source
This is a newsfeed from leading technology publications. No additional editorial review has been performed before posting.