Technology

Centre for Sectoral Economic Performance

Research Centre

Knowledge Categories

Economics & Public Policy

Academic area

Cyber strategy: bringing policy and academia together to boost the UK cyber sector

Main image: Newman Studio / iStock via Getty Images

3 minute read

Off

How AI avatars are revolutionising the student experience at Imperial Business School

Nai Li — Wed, 15 Jan 2025 12:14:02 +0000

How AI avatars are revolutionising the student experience at Imperial Business School

Nai Li Wed, 15/01/2025 - 12:14

More blogs

13 April 2026 Knowledge

Expert Q&A: Henry Dimbleby on the hidden ways ultra-processed foods harm our health and why the government must act now

20 March 2026 Knowledge

'Boomerasking': How to break the annoying habit hurting your workplace relationships

12 March 2026 Knowledge

Who decided what ambition looks like?

18 August 2025

As GenAI continues to reshape the educational landscape, business schools are exploring new ways to deliver teaching that is both scalable and human-centred. One promising development is the use of AI-powered faculty avatars, digital twins, that can support student learning while preserving the relational core of education. At the crossroads of projected growth in the AI Avatar market and ethical debates, we are beginning to recognise their potential to enhance the learning experience.

In January 2024, the IDEA Lab at Imperial Business School launched a small-scale pilot project in collaboration with David Shrier, Professor of Practice (AI & Innovation), developing an AI avatar named Davatar as an experimental element of his AI Ventures module. The pilot was designed to give ��ý hands-on experience with applied AI in a business context. Feedback was overwhelmingly positive, with many ��ý identifying Davatar as the standout feature of the module.

Building on this success, in spring 2025, we extended the use of AI avatars to support teaching modules like marketing management, entrepreneurship and innovation, and organisational behaviour on the Global Online MBA and Executive MBA programmes, and piloted three more avatars, MarkBot, OmarBot, and BartBot, which were developed in collaboration with the Touchcast Mentor Platform and Dr Ayelet Segal.

These avatars were not intended to replace human educators, but to supplement and enhance faculty presence in online and blended learning environments. Initial feedback suggests they have the potential to transform the way we support ��ý across different time zones, schedules, and diverse learning needs.

Enhancing faculty presence through digital twins

One of the ongoing challenges in online learning is the reduced sense of human connection. As per an article in Theory of Transactional Distance, ��ý often feel isolated, especially when studying asynchronously or across different time zones, and their success on the programme can be impacted by the degree of “transactional distance” between themselves and their teachers. The AI avatars were created to address this by replicating the voice, tone, and visual likeness of real faculty members, effectively becoming “digital twins.”

This was significant. In feedback surveys, ��ý consistently reported that it felt more natural and reassuring to interact with an avatar that resembled their lecturer. As one student said, “It’s him, just in another form. It makes it feel more like he’s here with us.” This was especially important during periods of high stress, such as when approaching assignment deadlines, when ��ý often seek last-minute reassurance or clarification.

By embedding academic identity into the avatar experience, we were able to foster trust and sustain the continuity of the learning relationship. Our research in this area will continue to explore whether digital twins can contribute to a reduction in the perceived transactional distance for ��ý in a blended or fully online programme.

Supporting flexible and on-demand learning

For ��ý juggling work, family, and studies, often across different time zones, the traditional model of office hours and scheduled tutorials can be limiting. One of the main benefits of the avatars was their 24/7 availability. Students could engage with them whenever they needed support, whether during a lunch break in Singapore or late at night in London.

Usage data from March to June 2025 showed clear spikes in avatar engagement just before major assignment deadlines. Students used them not for casual conversation, but for structured academic queries, whether seeking help with frameworks, clarification of learning outcomes, or guidance on assessments. This on-demand support reduced barriers to learning and made academic support more equitable and responsive.

In 1984, educational psychologist Benjamin Bloom reported that ��ý tutored one-to-one performed two standard deviations better than peers taught only in a group environment. Our research may help us understand whether AI avatars might one day help us solve the problem of scaling personalised one-to-one tutoring to make it available on-demand for all ��ý.

Strategic implications for educators and institutions

The pilot raises broader questions about how AI can be integrated into teaching practice. First, any introduction of AI must begin with pedagogy, not technology. It is Imperial Business School’s belief that tools like avatars should be designed to enhance – never to weaken or replace - the thoughtful dialogue and interaction between student and teacher that underpins effective learning.

Secondly, the identity of the educator is important. Avatars that resemble real faculty members, both visually and vocally, help build stronger bonds with ��ý. These aren’t anonymous chatbots; they are extensions of the educator’s presence. This sense of familiarity encourages engagement and helps ��ý see AI as an ally, not a threat.

Third, professional development will be crucial. To ensure faculty feel confident using these tools, we must offer opportunities to improve both technical skills and pedagogical understanding. This isn’t just about learning to operate a new platform; it’s about rethinking how we design and deliver teaching in an AI-enhanced environment.

Where do we go from here?

This was an initial pilot phase involving three modules and a small group of early-adopting faculty. The next step is to expand the initiative more broadly, testing AI avatar use across various disciplines, delivery modes, and levels of AI familiarity. In parallel, we will continue to research student and academic attitudes towards the use of AI tools in education, and to explore the ethical and environmental aspects of this technology.

Ultimately, our goal is not to replace the human elements of teaching, but to explore the responsible use of AI to enhance them. Digital twins offer a way to extend faculty presence, provide responsive support, and foster more inclusive learning experiences. When used thoughtfully, they may help us reimagine what personalised education can be like in an increasingly digital world.

Moving beyond FAQs: designing for dialogue

Many AI tools in education often fall short because they are designed to deliver content passively, primarily answering frequently asked questions, and do little to support critical thinking or deeper understanding. Our aim was different. We wanted to explore whether AI avatars could play a meaningful pedagogical role by facilitating a more interactive and reflective learning process.

To achieve this, we designed the avatar conversations based on Laurillard’s Conversational Framework, a model that sees learning as a dialogue between teacher and student. Students weren’t simply given answers. Instead, the avatars guided them back to key programme concepts and helped them explore different interpretations and applications.

For example, a student might ask, “Is this a good example of emotional intelligence in the team task?” or “Should I apply Porter’s Five Forces to the group case or the individual one?” Rather than giving binary responses, the avatars encouraged the student to reflect and engage more deeply with the material.

Initial feedback [on the AI-powered faculty avatars] suggests they have the potential to transform the way we support ��ý across different time zones, schedules, and diverse learning needs.

Discover more

IDEA Lab

Student Experience

Knowledge Categories

Academic area

Analytics & Operations

These AI-powered digital twins are transforming learning across time zones and diverse needs

Main image: OmarBot | Imperial Business School

5 minute read

Off

AI-powered rental fashion: Smarter recommendations for a sustainable future

Gah-Yi Ban — Wed, 15 Jan 2025 12:14:02 +0000

AI-powered rental fashion: Smarter recommendations for a sustainable future

Gah-Yi Ban Wed, 15/01/2025 - 12:14

More blogs

13 April 2026 Knowledge

Expert Q&A: Henry Dimbleby on the hidden ways ultra-processed foods harm our health and why the government must act now

20 March 2026 Knowledge

'Boomerasking': How to break the annoying habit hurting your workplace relationships

12 March 2026 Knowledge

Who decided what ambition looks like?

6 May 2025

Fashion is one of the most polluting industries in modern times. It produces about 10 per cent of the UK's annual carbon footprint – more than all international flights and maritime shipping combined. Research from Climatetrade, 2023, shows it also consumes vast amounts of water and generates millions of tons of waste. According to Fashion Transparency Index 2023, the constant demand for cheap clothing has made the industry a "key driver of human rights abuses". One of the largest study into clothing habits ever undertaken by climate action NGO WRAP shows the UK's wardrobes hold 1.6 billion items of unworn clothes.

To address some of these challenges, a new business model of subscription rental fashion has sprung up in recent years. Established brands like LK Bennett and Ann Taylor, along with startups such as Hurr, My Wardrobe HQ, nuuly, and Armoire, now offer rental services that allow multiple customers to “share” a much larger wardrobe, reducing the need for constant new purchases. These companies operate on a monthly subscription model, enabling members to rent and swap garments within each period. Returned garments are inspected, cleaned, and recirculated, ensuring efficient inventory use. The rental fashion business model has potential for reducing waste, lowering emissions, and extending the lifecycle of clothing, all the while keeping customers stylish and satisfied.

The challenge of keeping customers engaged

As operations researchers, my colleagues and I, together with a PhD student, Jiannan Xu, were drawn to rental fashion because of the operational complexity it presents. We partnered with a rental fashion startup and analysed its customer data. One issue stood out from the analysis: customers were not staying on the platform long enough. Many left after the initial discounted trial period, citing dissatisfaction with the style or fit of their rented items.

To improve retention and enhance the customer experience, we developed a novel approach to optimising personalised recommendations for rental fashion businesses. Our work focused on two key challenges:

Encouraging customers to subscribe – many companies offer discounted trials, yet potential subscribers hesitate because they’re unsure whether they’ll find styles they like.
Retaining customers beyond the trial period – a large percentage of cancellations stem from product fit uncertainty. Customers may select items based on images and descriptions, only to find that they don’t meet expectations in person. This mismatch leads to frustration, increased returns, and lower renewal rates.

Smarter assortment optimisation with AI

To tackle these issues, we developed an AI-powered algorithm that improves how rental companies curate product recommendations. Instead of relying on generic recommendation engines or brute-force optimisation, our model develops and optimises a stochastic decision model to account for the unique structure of rental fashion businesses. Specifically, it considers:

Initial selection uncertainty – predicting which items are most likely to encourage customers to subscribe.
Renewal uncertainty – identifying which items will maximise customer retention after the first rental cycle.

By carefully balancing these factors and using smart mathematical approximations, we developed a fast, scalable, and personalized ranking algorithm—one of the simplest and quickest methods for optimising the assortment curation process. Essentially, our algorithm gives a personalised score to each garment, whereby the score weighs in the expected consumer surplus (“signal”) and the uncertainties regarding the customer’s initial and post-wear valuations (“noise”). It thus has parallels with finance and engineering applications where good decisions are made by balancing signal-to-noise ratios.

Alternative methods to our algorithm would be either too slow for real-time use, or ineffective as they do not optimise the business objective. In extensive numerical simulations, we show our algorithm ranks and refines product assortments in milliseconds, compared to minutes using brute-force optimisation. We further show that, even under stress-testing scenarios, our algorithm can achieve results within 6.5 per cent of the best possible profit outcomes, making it both effective and scalable.

Practical benefits for rental fashion companies

Beyond technical advancements, our research has real-world implications for the rental fashion industry. By ensuring that customers receive better-matched recommendations, businesses can reduce unnecessary returns and improve overall satisfaction. This leads to:

Fewer shipments and exchanges, reducing logistics costs.
Better use of inventory, maximising item circulation.
Higher customer retention rates, strengthening long-term profitability.

The future of AI in subscription-based rentals

Beyond rental fashion, the principles behind our AI-powered optimisation model extend to subscription-based rentals of experience goods in general. This includes electronics rentals, furniture subscriptions, and toy libraries, where selecting the right assortment is key to customer satisfaction and retention. By refining product recommendations, businesses can optimise customer engagement, reduce churn, and improve profitability.

Our research is among the first to explore assortment optimisation in this space, setting the stage for further studies in inventory and subscription management. While rental businesses may not solve all the environmental challenges, smarter, AI-powered recommendations can significantly enhance business viability and the overall customer experience—paving the way for a smarter, more efficient future in rental commerce.

Explore key findings

Read the research from Gah-Yi Ban

Sustainability and climate change

Knowledge Categories

Entrepreneurship & Innovation

Management & Entrepreneurship

Academic area

Can AI optimise rental fashion sustainably while boosting retention?

Main image: alfexe / iStock via Getty Images Plus

4 minute read

Off

Profile: Associate Dean of Education Quality Cloda Jenkins on why Higher Education must embrace generative AI

Evie Burrows-Taylor — Thu, 05 Dec 2024 11:37:12 +0000

Profile: Associate Dean of Education Quality Cloda Jenkins on why Higher Education must embrace generative AI

Evie Burrows-Taylor Thu, 05/12/2024 - 11:37

13 April 2026 Knowledge

Expert Q&A: Henry Dimbleby on the hidden ways ultra-processed foods harm our health and why the government must act now

20 March 2026 Knowledge

'Boomerasking': How to break the annoying habit hurting your workplace relationships

12 March 2026 Knowledge

Who decided what ambition looks like?

Engaging with AI in education

“The biggest risk is deciding generative AI will undermine education or devalue what we do,” Dr Jenkins says, advocating instead for engaging with the technology to understand where it can be beneficial – and the ways in which its use should be limited.

Recently, Dr Jenkins worked with the Business School's IDEA Lab on stress testing module assessments across the Business School using a range of AI tools onto gain insights into limiting AI interference. The experiment looked at what happened if ��ý asked AI to complete module assessment tasks, with testers evaluating each task for accuracy, clarity, relevance, compliance, referencing and ease of use. Dr Jenkins reviewed the results along with the IDEA Lab’s learning designers, calculating average scores and assigning risk levels.

With her primary focus on ensuring the delivery of high-quality education programmes across the School, Dr Jenkins is keen to support both ��ý and faculty in the use of AI tools. But, she notes, the biggest challenge, particularly for ��ý, is in understanding where to draw the red lines.

“Mechanisms have to be put in place so that ��ý understand what is and isn’t allowed. For example, if we want to use generative AI for idea generation, we can facilitate that by making time for using AI tools in the classroom where there can be oversight,” she explains.

Preparing ��ý for the job market

There are also wider implications for incorporating generative AI into curriculums. For example, when it comes to future employability, an area Dr Jenkins has researched for a number of years. While knowledge of AI tools could help ��ý to develop necessary skills to navigate the job market, they will still have to be able to “critically assess, explain their ideas, and think on their feet”, which she believes is vitally important in preparing them not just for job interviews, but for life in general.

Dr Jenkins is familiar with what recruiters look for and how skill requirements evolve over time. Previously, she worked as a consultant and was headhunted to the independent UK energy regulator Ofgem to run its review of energy network regulation. As well as making use of her PhD on the regulation of price controls in the energy sector, during her time there she was exposed to strategic thinking around tackling big challenges.

The biggest risk is deciding generative AI will undermine education or devalue what we do

“This was an exciting time for me, as it was an opportunity to apply my PhD research and really shape policymaking,” she says, regarding the price control framework she designed and implemented, which is still in use today. “It was a very rewarding time in terms of producing something that actually had an effect."

As an expert in economics education, including on the connections between degrees and skill-development, one of Dr Jenkins' key interests is research-based education, with ��ý learning by doing. Her interest in this area stems from the criticisms levelled against economics for its reliance on theory that, at key moments such as the financial crisis and the COVID pandemic, hasn't always been effective.

“I’m very keen on education not just being about learning theory, but giving ��ý the tools they need to apply these theories to the real world,” she says, adding that research-based education encourages ��ý to explore questions and work out how to find the answers and methodologies that apply. “It’s about enabling ��ý to explore questions and feel comfortable with the idea that there isn’t always a right answer," she says.

Active learning

In fact, it was this active learning approach that attracted her to teaching at Imperial. “At the Business School, active learning and application is always front and centre and for me that’s really key as it better prepares ��ý for going into jobs in either the private or public sector.”

In addition to Dr Jenkins's roles as Associate Dean and co-running the Centre for Teaching & Learning Economics, she teaches five modules across BSc, MSc and MBA programmes – undoubtedly a lot to balance. But Dr Jenkins is unequivocal about the positions she holds at the School: “Being close to the ground is important to me. Fundamentally, I really enjoy teaching, but it also helps when I’m talking to colleagues in my capacity as Associate Dean – it gives me the chance to see both sides”.

Mechanisms have to be put in place so that ��ý understand what is and isn’t allowed

Whether teaching, considering how to deliver a new education programme, or implementing a new regulatory framework, Dr Jenkins is motivated by the desire to facilitate change and improve on how things are done.

“I’m focused on thinking about better ways of doing things. Everything I do is about finding the best path, whether in a regulation space or working with a team on a process at the Business School,” she says. “We live in a constrained world with constrained time, so thinking smarter about how we can do things is the way forward”.

Education and Learning

IDEA Lab

Leadership

Knowledge Categories

Economics & Public Policy

Academic area

As the higher education sector grapples with the role generative AI will play in the learning experience, the Business School's Associate Dean of Education Quality stresses the positives of this new technology.

4 minute read

How AI could use language to shape our reality

Mark Kennedy — Mon, 18 Nov 2024 10:51:23 +0000

How AI could use language to shape our reality

Mark Kennedy Mon, 18/11/2024 - 10:51

18 November 2024

Throughout human history, we have used language to define our experience of the world. But what happens if AI develops this capability?

Language reflects how we live and what we do. "Influencer" and "selfie" are terms most of us understand, but, go back 30 years, and nobody would know what they meant. They describe concepts that did not quite fit into any existing language that preceded them, and our adoption of them is a shared acceptance that they’re different enough and important enough to be worth learning.

These concept terms are known as categories. For most of history, the creation and dissemination of new categories has been a solely human activity. In recent years, artificial intelligence (AI) has evolved to such a point that it is now fully capable of doing the latter – teaching us about categories we, as individuals, may be unfamiliar with.

But will AI someday be capable of the former too? Will it be able to propose that there are things that we, as human beings, have not yet said or even appreciated about the world, and create new language that allows us to see reality in a different way? Our research explores this question.

Categories shape our reality

Category work is one of our apex functions as human beings. Throughout our history, the development and sharing of knowledge – which involves inventing labels for patterns and things we experience in the world – is something that has in itself changed the world. In other words, category creation is not just reactive – it actively shapes reality.

For example, we set the Industrial Revolution in motion, we learned how to shape a new world by burning fossil fuels at scale to generate power, and now we have global warming because this has put too much greenhouse gas into the atmosphere. All of this is the process of category creation.

We may... see AI move into other things we currently consider deeply human

Innovators and creatives are constantly engaging in this process, trying to change the world – and they have already started working with AI. Right now, AI is a tool that can help them organise their ideas and plans. But it’s arguably capable of participating more actively too.

Previous research has found AI capable of participating in the game of categories, so it’s logical to assume it will continue this journey to become an equal collaborator in future and – beyond that – an innovator in its own right. This would see AI join human innovators as a creative partner, interacting with them like a real person and performing traditionally human roles.

We are still a long way from this potential future, and it’s impossible to know now what categories or types of categories AI may produce down the line. If we could know what they are, they would already exist. But we can take a guess, based on the aforementioned trajectory of AI from tool, to participant, to collaborator, to innovator.

Virtual therapists

In his book How To Know a Person: The Art of Seeing Others Deeply and Being Deeply Seen, David Brooks argues we as humans are not as good at truly seeing one another as we like to think. The key, according to Brooks, is to shine a positive light on others by taking a genuine interest and really listening to them. Doing this helps us to really know a person and, in the process of being known, helps the other person to better know themselves.

Most people are not very good at this. It takes attention, full engagement, a willingness to accompany someone, and the ability to engage. But what if generative AI were able to interact with people in that way? What if it could help us see ourselves in a new or better way?

Category creation is not just reactive – it actively shapes reality

Most people right now would baulk at the idea of a virtual therapist, but at the same time many institutions – universities, for example – lack sufficient therapy resources to support the scale of mental health challenges they face. With this in mind, a first-line option of an AI therapist might not be a bad idea.

And, if this were possible, we may also see AI move into other things we currently consider deeply human and utterly removed from the sphere of computing. For example, caring and nurturing roles that currently require a long time to learn, a great deal of intelligence and expensive resources, but could be done by AI at lower cost and greater scale.

This isn’t a direct prediction, but an area of work that is perhaps ripe for development. In such a future, a supportive AI may be able to alert us to problems, issues and improvements to our lives that we cannot currently even conceive of. This, in essence, will involve the creation of new categories that do not exist today, but which will quickly become part of our social landscape, heralding a strange yet exciting world to come.

This article draws on findings from "Beyond the Turing Test: Exploring the Implications of Generative AI for Category Construction" by Nelson Phillips (University of California), Sai Kalvapalle (Nova School of Business and Economics) and Mark Kennedy (��ý).

Jobs and Labour

Monthly newsletter

Receive the latest insights from Imperial Business School

Expert Q&A: Is AI going to save us or kill us?

World Economic Forum Fellow and Visiting Lecturer at Imperial Business School Dr Catherine Mulligan explores what AI means for society.

Management & Entrepreneurship

Academic area

Knowledge Categories

Expert Q&A: Digital transformation in action at Vodafone

Evie Burrows-Taylor — Thu, 12 Sep 2024 10:02:42 +0000

Expert Q&A: Digital transformation in action at Vodafone

Evie Burrows-Taylor Thu, 12/09/2024 - 10:02

12 September 2024

Professor of Practice David Shrier sits down with Vodafone Business UK's Chief Data Officer & Head of Portfolio Myriam Salah to discuss the company's digital transformation journey.

DS: To start, can you describe your role within Vodafone?

MS: My role at Vodafone focuses on digital data and transformation. My main responsibility is to ensure that our organisation becomes a data-driven entity. This involves not only having exceptional salespeople and account teams but also empowering them with accurate and timely data. A couple of years ago, we decided to create a state-of-the-art data team, appoint a Chief Data Officer, and drive our journey towards becoming a data-literate organisation. This transformation is about making data readily available, democratising its access, and training our people to use it effectively.

DS: Can you elaborate on how you're approaching digital transformation at Vodafone?

MS: Digital transformation at Vodafone is a multifaceted endeavour. First, we aim to offer our customers the option to do everything unassisted if they choose, with the ability to seamlessly transition to assisted support as needed. To achieve this, we need a robust data layer that connects all customer-facing interfaces with our internal systems. This ensures a cohesive and integrated experience.

Additionally, we are focused on consolidating the various systems we have acquired over the years through mergers and acquisitions. By integrating these systems, we can provide an omnichannel experience, run analytics on a centralised data cloud, and ultimately support our corporate strategy of putting customers first and simplifying our business processes.

DS: It sounds like a comprehensive strategy. Can you describe how your team is organized to support these initiatives?

MS: Our team is structured into several key areas. In the data space, we have teams focused on data movement and transformation, working with our technology colleagues to build data pipelines and move data into the cloud. Our data science and insights team is split into two areas: insights, which build statistical reports for day-to-day operations, and data science and machine learning engineering, which develop and deploy machine learning models.

On the digital transformation side, we have an e-commerce team, an omnichannel team, and a trade trading and digital advocacy team. Each of these teams focuses on different aspects of our digital strategy, from web applications and mobile apps to ensuring all our software platforms communicate effectively and drive digital advocacy internally.

DS: What are some of Vodafone's key objectives for 2024?

MS: For 2024, we have several key objectives. One major goal is to increase traffic to our online interfaces by 20 per cent, justifying further investments in our digital platforms. We also aim to consolidate our various portals into a unified system, enhancing user experience and efficiency. Additionally, we want to be recognised as a digital-first brand, which is crucial for our customer interactions and overall satisfaction.

DS: How is Vodafone leveraging AI in its digital transformation efforts?

MS: AI is a significant part of our strategy, though we approach it with caution due to regulatory and ethical considerations. We're developing AI use cases, particularly in the area of generative AI, to handle the vast amounts of text data across our business. For instance, we're working on a ChatGPT-like search functionality to help our employees quickly find information and resolve customer issues. This will enhance customer service and operational efficiency.

DS: Beyond AI, what other digital trends and technologies are you tracking closely?

MS: One trend we're closely monitoring is the concept of total experience, which aims to provide a seamless and consistent experience across all digital interactions for both customers and employees. We're also focused on the potential of 5G standalone deployment, which we believe will transform how we offer services and build modular product offerings.

DS: Looking ahead, what do you think will be the biggest disruptor for Vodafone's business in the next few years?

MS: We're keeping a close eye on AI and its transformative potential, particularly how it might automate many tasks currently handled through traditional digital transformation initiatives. Additionally, sustainability is becoming a crucial differentiator, and we're exploring how technology can support sustainable development goals. Access to connectivity through 5G and modular product builds will also play a significant role in shaping our future offerings.

DS: Finally, what advice would you give to other organisations embarking on their digital transformation journeys?

MS: My advice would be to clearly define what digital transformation means for your organisation and ensure it aligns with your long-term business goals. Conduct a thorough impact assessment to understand where you stand and create a detailed plan. It's essential to get buy-in from all stakeholders, as digital transformation often involves significant changes in processes and technology. Breaking the transformation into smaller, manageable chunks that deliver quick wins can help maintain momentum and engagement.

DS: Myriam, thank you for sharing these valuable insights. It's clear that Vodafone is making significant strides in digital transformation, and I look forward to seeing how your initiatives unfold in the coming years.

MS: Thank you, David. It's been a pleasure discussing our journey and the exciting future ahead.

Imperial Business School's Centre for Digital Transformation is collaborating with Vodafone on investigating the impacts of digital technologies on business, and strategies Vodafone can adapt to its context as it navigates this new landscape. This interview has been published with Vodafone's permission.

Leadership

Monthly newsletter

Receive the latest insights from Imperial Business School

How AI could usher in a new age of scientific discovery

New research from Imperial Business School shows how AI could be used alongside traditional scientific methods to unlock the secrets of life, nature and the universe.

Management & Entrepreneurship

Academic area

Knowledge Categories

How AI could usher in a new age of scientific discovery

Ryan Cory-Wright — Mon, 19 Aug 2024 11:13:34 +0000

How AI could usher in a new age of scientific discovery

Ryan Cory-Wright Mon, 19/08/2024 - 11:13

20 August 2024

New research from Imperial Business School shows how AI could be used alongside traditional scientific methods to unlock the secrets of life, nature and the universe.

As the prominence and capability of AI (artificial intelligence) has increased, so has AI-related anxiety. While much of this focuses on generative AI and the potential for it to replace human jobs, the existential fear often extends to all forms of AI.

However, new research from Imperial Business School, produced with colleagues at IBM and Samsung, has identified the valuable – possibly critical – role AI can play in furthering human understanding about nature, the world and the wider universe.

Specifically, our research shows how we can work with AI and traditional scientific approaches to discover new laws of nature, which are provable, robust, and consistent with existing laws. This simultaneously addresses the drawbacks of the traditional scientific method, as well as a purely data-driven approach.

The traditional scientific method

Since the 17th century, the scientific method has been the key driver for human discovery of the laws of nature. The approach sees researchers create hypotheses using both existing theory and observations, and then test those hypotheses through experiments.

However, the rate at which this method yields discoveries has been steadily dropping, with academics observing a decline in research productivity that suggests ideas are simply becoming harder to find.

AI-Hilbert was able to successfully derive Kepler’s Third Law of Planetary Motion and Einstein’s Law of Relativistic Time Dilation

Physicist Paul Dirac summarised the issue neatly when he observed that, “in the early days of quantum mechanics a second-rate physicist could do first-rate work, [but] it has since become difficult for a first-rate physicist to do second-rate work.”

As Dirac implies, it seems that much of the low-hanging fruit of scientific discovery has now been picked, leaving more challenging, work-intensive problems to be solved. With this in mind, if we do not want to accept the gradual decline of human discovery and innovation, it’s clear a solution is needed.

Boosting evidence with data

Fortunately, alongside this decline we have also seen rapid technological advances in terms of data processing; researchers estimate that raw computing power increased at least six orders of magnitude from 1991 to 2015. This has led to data-driven approaches emerging as viable paths to new scientific discoveries, with AI tools able to be deployed to identify patterns in large datasets.

In their current form, however, these tools often fail to provide formal proofs for discovered relations or hypotheses, and are prone to "hallucinating" incorrect scientific laws that cannot be either proven or disproven. That means new laws of nature discovered in this way are difficult to explain or justify, making it an insufficient substitute for the traditional scientific method.

Much of the low-hanging fruit of scientific discovery has now been picked

Our research addresses this problem, setting out a new approach to scientific discovery that brings together existing theory and data each step of the way. This could be the first step towards a scalable solution, as it provides a means of identifying scientific laws from large datasets while providing proof of their consistency with existing theory.

This will allow new discoveries to be made in areas where background theory is limited, and in areas where gathering data is prohibitively expensive.

Testing the data-driven approach

To test our new approach – named AI-Hilbert, after the mathematician David Hilbert – we used it to derive some of the most frequently cited laws in scientific literature. AI-Hilbert was able to successfully derive Kepler’s Third Law of Planetary Motion and Einstein’s Law of Relativistic Time Dilation, among others.

Breaking this down further to check different scenarios, we tested whether AI-Hilbert was able to derive valid scientific laws purely from complete and consistent background theory. For this, we used it to derive the Hagen-Poiseuille equation.

We then tested whether it was also capable of deriving laws from inconsistent background theory. AI-Hilbert successfully overcame this by employing data to identify the correct theory, highlighting one of the key benefits of the unified approach. We also found that the system is data-efficient, meaning that as more background theory is provided, it requires less data to derive a scientific law.

A new age of discovery

By demonstrating that an AI system can be made capable of rediscovering existing scientific laws afresh, while dealing with contradiction, we show that it is also likely to be capable of identifying new scientific laws in the future. This potentially ushers in a new AI-driven age of discovery to rival the golden age of the scientific method.

This highlights a useful, supportive and non-attritional role for AI in helping us grow and develop as a species. Only time will tell whether this comes to pass, but by providing the tools and evidence of their potential, we hope that our study brings us one step closer to unravelling some of the mysteries of the universe.

This article draws on findings from "Evolving scientific discovery by unifying data and background knowledge with AI Hilbert", by Ryan Cory-Wright (��ý), Cristina Cornelio (Samsung AI), and Sanjeeb Dash, Bachir El Khadir and Lior Horesh from the IBM Thomas J. Watson Research Center.

Education and Learning

Monthly newsletter

Receive the latest insights from Imperial Business School

How is AI affecting freelance jobs?

Which industries is generative artificial intelligence reshaping? Who are the winners and losers of the automation of expertise? What impact will this have on livelihoods and economies?

Academic area

Analytics & Operations

Knowledge Categories

How data can boost charity fundraising

Evie Burrows-Taylor — Tue, 26 Mar 2024 11:36:47 +0000

How data can boost charity fundraising

Evie Burrows-Taylor Tue, 26/03/2024 - 11:36

26 March 2024

Charities need more sophisticated ways of fundraising – and data analytics is key.

Fundraising is an expensive and time-consuming business for charities. Many potential donors are unlikely to respond to digital appeals, yet mail and phone campaigns can be costly and inefficient. Without more sophisticated analysis, how can fundraisers know whether last year’s donor is likely to contribute again this year, where to find new donors, or how often to ask for money without appearing greedy?

There is a better way, say Imperial Business School’s Dr Sven Mikolon and Professor Wolfram Wiesemann, who have used data analytics and artificial intelligence (AI) to provide insights into how charities and nonprofit organisations can raise cash – with dramatic improvement.

“Charities are advanced in their conceptual approach to campaigning but there’s a lot of money they leave on the table,” says Dr Mikolon. “The potential of key operations – donor targeting, campaigning, and retaining donors – isn’t fully exploited.”

Efficient fundraising

Using data from a charity partner, their team trained and tested AI to target donors more efficiently during fundraising campaigns. Results showed a 12 per cent increase a year in profits – gains worth £500,000 a year to the organisation, which supports children’s welfare and families. “We used data to train an algorithm to assess whether the cost of sending a letter, for example, would outweigh the likely donation they would receive from that individual. Then we tested [the AI] on data it hadn’t seen before to understand how it performed.”

The team will soon test their technology on a live fundraising campaign. This will also allow the organisation to understand how individuals might respond when they are no longer contacted.

Charities are advanced in their conceptual approach to campaigning but there’s a lot of money they leave on the table

When following up on fundraising campaigns, charities need to know information such as how often individuals have been contacted, how much they tend to give, which campaigns they respond to – and this information is tricky to gather during traditional campaigns.

Unlike “black box” AI, where internal workings are hidden to users, the analytics used by Imperial will be transparent and accountable. “It’s important for us that we use explainable AI,” says Dr Mikolon, “so we can show exactly how a decision is made to include a donor in a campaign.”

Understanding the data

If rolled out more widely, fundraising analytics could bring sizeable improvements to a sector that has lagged behind other industries in its adoption of technology, says Professor Wiesemann. But the potential is great: in the UK alone, the nonprofit sector contributes around £15 billion to GDP and close to 900,000 jobs.

“As long as you have data available to train predictive models, that will allow you to come up with these predictions of expected donations,” says Dr Mikolon.

Imperial’s expertise allowed the project to proceed smoothly, says the head of data science for the charity partner, and has helped raise awareness of the potential of AI within the organisation: “The strategy proposed by Imperial will be highly relevant in enhancing the efficiency of direct mail campaigns, which remain vital to us.”

Benefits are filtering through to the Business School too: postgraduate ��ý who have worked with Dr Mikolon and Professor Wiesemann on a variety of linked projects, including understanding data sets provided by charities before the data can be leveraged by AI.

“I think it really enriches the experience at Imperial,” says one MSc student. Another relished the chance for hands-on experience: “I’ve gained exposure to many valuable concepts which I’m already using for my new job.”

by Helena Pozniak

Competition

Monthly newsletter

Receive the latest insights from Imperial Business School

How AI and XR can transform education and create a new dimension of learning

Artificial intelligence and extended reality can create new ways of learning that are more engaging, personalised and human-centric. By using them, we can solve some of the biggest challenges in education.

Academic area

Analytics & Operations

Knowledge Categories

How is AI affecting freelance jobs?

Xinrong Zhu — Wed, 24 Jan 2024 10:37:19 +0000

How is AI affecting freelance jobs?

Xinrong Zhu Wed, 24/01/2024 - 10:37

5 February 2024

Which industries is generative AI reshaping? Who are the winners and losers of the automation of expertise? What impact will this have on livelihoods and economies?

Early in 2023, we learned that freelance writers in Kenya were struggling to make a living in a once thriving market. For years, remote ghostwriters have supported themselves and their families by writing essays and other tasks for US-based ��ý.

But within a few months of the launch of ChatGPT, these freelance workers have seen demand for their skills plummet, along with their pay, as ��ý turned instead to AI to do the job they’d once contracted to humans.

Since ChatGPT launched in November 2022, its use has soared. We wanted to know who else was feeling the brunt of automation beyond freelance writers.

Our initial investigations reveal that the rise of new generative AI tools has been swiftly followed by a substantial fall in demand for freelancers. Seven months after ChatGPT’s release, online freelancers in professions that are more vulnerable to automation saw an overall 21 per cent fall in weekly demand for their skills compared to those whose jobs involve more manual tasks. Results are clear: generative AI is reshaping job markets.

How did we discover which jobs are being affected by generative AI?

A vast international freelancing platform tells the story through the job posts of the employers who use it. Thousands of new jobs are advertised on this site daily – it’s a reflection of the fast-changing worldwide market for short term, task-based remote work.

With access to two years’ worth of data from July 2021 to July 2023, we analysed nearly two million (1,725,587) unique job posts from all over the world. We were able to examine in depth how the volume and type of work in demand changed after the introduction of ChatGPT and other AI tools.

Previously, when researchers have assessed the impact of AI on jobs, they’ve concentrated on writing. But we’ve broadened this investigation to include a range of exposed professions, including work that involves coding and automation, such as engineering, software, and app and web development. These are the types of expertise that new generative AI is becoming increasingly adept at replacing.

Online jobs less affected by ChatGPT include more manual work such as data entry, video and audio editing, and social media post production. But visual creative professions aren’t immune to automation, our analysis found: the advent of text-to-image generation through AI has seen work in graphic design and rendering of 3D images experience a 13 per cent fall in demand.

Online freelancers in professions that are more vulnerable to automation saw an overall 21 per cent fall in weekly demand for their skills

The 21 per cent difference between jobs more and less prone to automation is an accurate reflection of the demand from employers on the freelance platform. Previous research has focused upon the supply side (the jobs accepted by freelancers) but this statistic might be affected by other factors such as quality and freelancers’ skills.

Interestingly, we discovered that, rather than falling, rates of pay rose very slightly for the jobs prone to automation. Our hunch is that it’s the simpler, shorter-term jobs that are being replaced by AI, whereas employers still require human expertise for the more complex tasks – which are more valued by employers and where freelancers can demand higher rates. While there’s no direct means to measure job complexity, we were able to estimate this from the job descriptions themselves.

We backed up these findings with a couple of separate investigations. Patterns from Google searches confirmed that job categories linked to higher awareness of generative AI also exhibit a more significant decline in demand from employers. In addition, a metric which measures the exposure of certain types of work to artificial intelligence also confirmed our findings.

What will the future impact of generative AI be?

Our work raises questions for future research. Is the impact of generative AI more pronounced in freelance markets than in traditional corporate environments? Online work is typically short-term, tightly defined and transactional; it’s a flexible market where suppliers (the freelancers) are less likely to have a relationship with employers and may live in a different country.

By contrast, traditional work environments might be more regulated and unionised, less flexible and more stable. Yet online freelance platforms play a significant role in economies: latest figures show they provided work for some 8.5 million freelancers worldwide, while some 2.3 million work full time across the range of platforms.

We’ve shown the difference between jobs more and less at risk of automation – but are the numbers of job postings falling overall? And among the professions more prone to automation, such as writing and coding, are some more vulnerable than others? Writing jobs, we found, fell by 30 per cent, whereas coding fell by 20 per cent. 

While our findings may paint a gloomy picture for online freelancers, as ever there will be winners as well as losers in this era of change. Whenever technology sweeps professions aside, new jobs emerge. New paradigms bring new opportunities and propel industries into unknown futures. Today, tasks that require greater degrees of creativity, intricate problem-solving and nuanced understanding remain the preserve of humans. The future will require different skills – and the human intelligence to provide them. 

Jobs and Labour

Monthly newsletter

Receive the latest insights from Imperial Business School

Expert Q&A: Is AI going to save us or kill us?

World Economic Forum Fellow and Visiting Lecturer at Imperial Business School Dr Catherine Mulligan explores what AI means for society.

Academic area

Marketing

Knowledge Categories

How AI and XR can transform education and create a new dimension of learning

Monica Arés — Mon, 15 Jan 2024 09:56:51 +0000

How AI and XR can transform education and create a new dimension of learning

Monica Arés Mon, 15/01/2024 - 09:56

15 January 2024

Artificial intelligence and extended reality can create new ways of learning that are more engaging, personalised and human-centric. By using these technologies, we can solve some of the biggest challenges in education

We are on the brink of a new era for both technology and education. At the forefront of this revolution is a convergence of AI and XR, causing a merger of the digital and physical world and offering us an entirely new set of tools. Through this combination, the Imperial IDEA Lab is pioneering future-focused solutions set to revolutionize global education and shape a future where technology and human potential unlock entirely new forms and dimensions of learning.

The evolution of learning: a historical perspective

In the space of one human lifetime, computers have gone from basic, simple machines to having the ability to analyse complex systems and predict outcomes.

Most of these technologies, however, have been largely passive, obediently doing what we instructed them to. Historically, the ways we teach and learn have mirrored this relationship, with rote memorisation and standardised testing being the norm. No one believed that was the best way to learn – it didn’t do much to foster skills such as creativity, critical thinking or collaboration – but we didn’t have the tools to make learning engaging at scale.

By using the power of these technologies, we can transform education from a passive, one-size-fits-all system to an active, customised and engaging one

Most of us can recall our favorite “aha moments” and the person, place or experience that inspired us and made us curious enough that we still remember them today.

The problem is that most of us only have a handful of these moments, making them the exception and not the norm. By not prioritising these true learning moments, we have inherited an education system that is critically flawed: it's a system that makes jobs susceptible to automation and leads to uninspired and disengaged educators and learners.

AI and XR as the ultimate tools for human potential

If we stay this course, prioritising a one-directional flow of content and information, we will be shaping our ��ý to be like robots – but now the robots are here, we can turn education’s focus onto being human.

This means focusing on what humans are actually good at, like awareness, higher-order thinking and decision making. Technology can then be used to complement these skills, by handling tasks that are tedious, time consuming, or beyond our capabilities.

What sets us apart from machines is our ability to feel curiosity, to imagine, and even our capacity to get frustrated when we can’t get it right in hopes of finding a better solution. This is what gives us the freedom to explore and innovate.

As humans, however, we are limited, and need to rely on the symbiotic relationship with technology and machines to help us to solve problems we have been unsuccessful at solving on our own.

No one believed that was the best way to learn [...] but we didn’t have the tools to make learning engaging at scale

We are starting to see this triangular partnership being used to build safer vehicles, cure diseases and tackle the climate crisis. In the same way that this hybrid combination will help us solve humanitarian and health issues, it can also help us solve some of the biggest challenges we face with education today.

The Imperial IDEA Lab is at the cutting edge of this work. By using the power of these technologies, we can transform education from a passive, one-size-fits-all system to an active, customised and engaging one. We can also unlock new dimensions of learning, where we can explore virtual worlds, manipulate digital objects, and interact with intelligent agents.

In my next article, I will dive deeper into the practical applications of AI and XR in learning and showcase some of the innovative projects and experiments we are working on at the Imperial IDEA Lab, and introduce you to the learning continuum.

Careers

Education and Learning

IDEA Lab

Leadership

Monthly newsletter

Receive the latest insights from Imperial Business School

Will AI replace the educator?

White-collar jobs are set to be transformed by AI – and business schools must adapt teaching practices quickly, so they can prepare ��ý for the new world of work.

Knowledge Categories