AI set to power ‘cognitive transformation’ of businesses say futures experts

Tech Foresight Day 2025

Academics and foresight experts joined a discussion with innovators from global companies about the future of AI at an Imperial Tech Foresight event.

With businesses still only part way through the digital transformation made possible by turn-of-the-millennium technologies such as cloud computing, AI is set to change everything again.

Experts from Imperial Tech Foresight joined researchers from Imperial and companies such as BP, IBM, PepsiCo and Unilever last week to explore 'cognitive transformation', the potential for AI to re-shape how organisations think and make decisions.

The importance of good data

Companies are already using AI not only to make office tasks more efficient but also to extract insights from large datasets, model business processes, and accelerate research and development.

The cognitive transformation is here. It’s more than just technology change; using new tools in turn changes the way that we think. The question is not whether businesses will use AI – but how wisely, ethically, and creatively they will do so. Nick Price Imperial Tech Foresight

One key application of AI is scientific discovery, as Robin Yellow, Principal in Digital Science and Engineering at BP, emphasised.

“Our scientists are using it to crack really difficult problems. Quantum chemistry for example, and chemical engineering problems such as unlocking biofuels.” He said that BP’s investment in high-performance computing underscores the strategic importance of AI in energy innovation.

Dr Tristin Brisbois, Senior Director of the Life Sciences Living Lab at PepsiCo, said that the company is using AI for applications such as optimising agriculture, understanding consumers, and helping nudge people to make healthy choices.

She added that new opportunities are becoming available as they better structure their company data. “In agriculture, we’ve been using AI for quite some time, because the data has been organised and data scientists have been well integrated. Elsewhere, we’ve had to organise our data and connect it to generate insights.”

“Sometimes we don’t have enough data, sometimes we have too much," said Dr Rossella Arcucci, Director of Research at Imperial’s Data Science Institute. We can get more out of our data, she said, using techniques her group is developing for data augmentation and reduction.

Making AI trustworthy

The adoption of AI is fraught with technical, ethical, and organisational challenges. Professor Francesca Toni in Imperial’s Department of Computing warned of the dangers of black box AI systems: “We don’t know what goes on inside. AI is very prone to biases because it’s trained on biased data.” She cited examples where identical individuals, differing only by gender, unfairly received different financial decisions from AI systems.

Imagine if everyone had an AI clone that interprets data in the blink of an eye. That’s an opportunity but also a danger. If you’ve got a bad business process, you’re going to produce a huge amount of rubbish in a short space of time. Robin Yellow

Professor Toni advocated for ‘computational argumentation’, a technique that enables AI systems to reason, explain, and be challenged. “We want AIs to be able to debate. We want humans to interact with AI and intervene in its reasoning – and vice versa.”

Data privacy and security are also major concerns. Dr Arcucci described how these could be partly addressed through ‘honest computing’ environments where sensitive data can be used to train models without being exposed: “You can run AI on the data, but you can never visualise or unload it.” However, concerns remain about whether black box models might still leak identifiable information.

Mr Yellow of BP warned that AI could risk propagating faulty business processes: “Imagine if everyone in this room had an AI clone. You bring your way of interpreting data to the task – but do it in the blink of an eye. That’s an opportunity but also a danger,” he said. “If you’ve got a bad business process, you’re going to produce a huge amount of rubbish in a short space of time.”

Top tips for harnessing AI effectively

The experts identified several approaches that could help organisations harness the full potential of AI while mitigating risks.

• Adopt AI agents
  Interest is increasing in AI agents, which are designed to carry out domain-specific tasks, often autonomously, and to interact with other agents and people. “In principle, with an agent you can get better performance out of a smaller model for a particular task. And because they’re more autonomous, you can end up with a workflow that you didn’t have to program,” said Dr Bradley Eck, Senior Research Scientist and Manager at IBM.
  
  In addition to improving performance, AI agents have the potential to be more trustworthy than general purpose AI models: “We’re creating AI agents with limited data access and training them to ensure the answers are correct,” said PepsiCo’s Dr Brisbois. Having multiple agents can enable greater diversity of AI thinking. “You don’t want a monoculture, but a diverse range of agents,” said BP's Mr Yellow.
• Favour explainable AI
  Professor Toni emphasised the need for AI systems whose decision-making procedures we can understand and intervene in. “When you look at many visions of ethical uses of AI there is a need for transparency,” she said. “We need to make sure we have explainability and the ability to contest AI decisions. This is not just for policymakers but also industry.”
• Invest in data infrastructure
  Echoing Dr Brisbois’s comments about data organisation, Dr Mike Dale, Digital R&D Director at Unilever, said that foundational data management unlocked innovation: “It seems old hat in the world of AI, but getting the organisation to ensure data quality and full scope was fantastic. That unleashed so much energy.”
• Foster deep partnerships with other organisations
  Collaboration between partners in business, government and academia is crucial, said Dr Dale. “What we’ve learnt is that you need intimacy with your partners and intensity – move away from treating them as contract partners and treat it as a longer-term partnership where you have a mutual interest in the outcome. We’re moving into an era where you need a technical partner, a science partner, and domain experts.”
• Educate and upskill your workforce
  AI literacy is essential. “Not everyone will be skilled, but upskilling as much as you can benefits all,” said Dr Brisbois. Dr Arcucci added, “We teach students not to avoid LLMs [large language models], but to understand their limitations and question their answers.”
• Improve governance and regulation
  Businesses need to empower employees to take advantage of AI responsibly, said PepsiCo's Dr Brisbois: “You can’t just say don’t use open AI – it’s a complete game changer. But you need guardrails. Tools need to get approved to ensure trade secrets aren’t flowing back into them.”
  
  Dr Arcucci called for stronger regulation. “What I wish for, if I can be completely honest, is more regulations covering the use of AI. Not for us but for next generations – that is what is really worrying me.” On a positive note, Dr Peter Pesl, Head of Innovation at ADVANTAGE AUSTRIA UK said: “The UK is very well positioned in ethical AI" due to the extensive research underway here on the topic.

The future of cognitive transformation

Looking forward, the integration of AI into business will deepen. Dr Brisbois predicted rapid advances. Reflecting on the future of advanced general intelligence (AGI), a more flexible and human-like intelligence that some commentators are mooting, she said: “AGI is just branding at the moment. But it is likely to be more than that in five to seven years.”

As AI becomes more embedded in workflows, the focus will shift from experimentation to execution. Nick Price, Foresight Strategy Manager at Imperial Tech Foresight, called for business leaders to “Think like a futurist, act like a strategist. Make small moves, not huge ones.” He added: “The cognitive transformation is here. It’s more than just technology change; using new tools in turn changes the way that we think. The question is not whether businesses will use AI – but how wisely, ethically, and creatively they will do so.”

Alex Elkins, Co-Director of Enterprise (Partnerships) at Imperial, said: “We’re delighted to be working in close collaboration with industry to pool expertise and resources, and take full advantage of AI’s positive transformative potential. Imperial has one of the largest concentrations of world-class AI researchers in the world, we work closely with industry to help inform how AI can offer positive transformative potential'.”

Tech Foresight Day is an annual events for businesses by Imperial Tech Foresight. The in-house futures practice is part of Imperial Business Partners, a corporate membership network that helps businesses access the best of Imperial's experts, talent and innovations.

--

Photos (from top-to-bottom) show the audience watching the panel discussion at Tech Foresight Day 2025; panellists Dr Peter Pesl, Robin Yellow, Dr Tristin Brisbois and Dr Rossella Arcucci; Professor Francesca Toni delivering a keynote; and a slideshow of participants in the day. Photos: Jo Mieszkowski / Imperial College London