Kay Firth-Butterfield is a globally recognised leader in ethical artificial intelligence and a distinguished AI ethics speaker. As the former Head of AI and Machine Learning at the World Economic Forum (WEF) and one of the foremost voices in AI governance, she has spent her career advocating for technology that enhances, rather than harms, society.
We spoke to Kay to discuss the promise and pitfalls of generative AI, the future of the Metaverse, and how organisations can prepare for a decade of unprecedented digital transformation.
Generative AI has captured global attention, but there’s still a great deal of misunderstanding around what it actually is. Could you walk us through what defines generative AI, how it works, and why it’s considered such a transformative evolution of artificial intelligence?
It’s very exciting because it represents the next iteration of artificial intelligence. What generative AI allows you to do is ask questions of the world’s data simply by typing a prompt. If we think back to science fiction, that’s essentially what we’ve always dreamed of — just being able to ask a computer a question and have it draw on all its knowledge to provide an answer.
How does it do that? Well, it predicts which word is likely to come next in a sequence. It does this by accessing enormous volumes of data. We refer to these as large language models. Essentially, the machine ‘reads’ — or at least accesses — all the data available on the open web. In some cases, and this is an area of legal contention, it also accesses IP-protected and copyrighted material. We can expect a great deal of legal debate in this space.
Once the model has ingested all this data, it begins to predict what word naturally follows another, enabling it to construct highly complex and nuanced responses. Anyone who has experimented with it knows that it can return some surprisingly eloquent and insightful content simply through this predictive capability.
Of course, sometimes it gets things wrong. In the AI community, we call this ‘hallucination’ — essentially, the system fabricates information. That’s a serious issue because in order to rely on AI-generated outputs, we need to reach a point where we can trust the responses. The problem is, once a hallucination enters the data pool, it can be repeated and reinforced by the model.
While much has been said about generative AI’s technical potential, what do you see as the most meaningful societal and business benefits it offers? And what challenges must we address to ensure these advantages are equitably realised?
AI is now accessible to everyone, and that’s incredibly powerful. It’s a hugely democratising tool. It means that small and medium-sized enterprises, which previously couldn’t afford to leverage AI, now can.
However, we also need to be aware that most of the world’s data is created in the United States first, followed by Europe and China. There are clear challenges regarding the datasets these large language models are trained on. They’re not truly using ‘global’ data. They’re working with a limited subset. That has led to discussions around digital colonisation, where content generated from American and European data is projected onto the rest of the world, with an implicit expectation that others will adopt and use it.
Different cultures, of course, require different responses. So, while there are countless benefits to generative AI, there are also significant challenges that we must address if we want to ensure fair and inclusive outcomes.
The Metaverse has seen both hype and hesitation in recent years. From your perspective, what is the current trajectory of the Metaverse, and how do you see its role evolving within business environments over the next five years?
It’s interesting. We went through a phase of huge excitement around the Metaverse, where everyone wanted to be involved. But now we’ve entered more of a Metaverse winter, or perhaps autumn, as it’s become clear just how difficult it is to create compelling content for these immersive spaces.
We’re seeing strong use cases in industrial applications, but we’re still far from achieving that Ready Player One vision — where we live, shop, buy property, and fully interact in 3D virtual environments. That’s largely because the level of compute power and creative resources needed to build truly immersive experiences is enormous.
In five years’ time, I think we’ll start to see the Metaverse delivering on more of its promises for business. Customers may enjoy exceptional shopping experiences—entering virtual stores rather than simply browsing online, where they can ‘feel’ fabrics virtually and make informed decisions in real time.
We may also see remote working evolve, where employees collaborate inside the Metaverse as if they were in the same room. One study found that younger workers often lack adequate supervision when working remotely. In a Metaverse setting, you could offer genuine, interactive supervision and mentorship. It may also help with fostering colleague relationships that are often missed in remote work settings.
Ultimately, the Metaverse removes physical constraints and offers new ways of working and interacting—but we’ll need balance. Many people may not want to spend all their time in fully immersive environments.
Looking ahead, which emerging technologies and AI-driven trends do you anticipate will have the most profound global impact over the next decade. And how should we be preparing for their implications, both economically and ethically?
That’s a great question. It’s a bit like pulling out a crystal ball. But without doubt, generative AI is one of the most significant shifts we’re seeing today. As the technology becomes more refined, it will increasingly power new AI applications through natural language interactions.
Natural Language Processing (NLP) is the AI term for the machine’s ability to understand and interpret human language. In the near future, only elite developers will need to code manually. The rest of us will interact with machines by typing or speaking requests. These systems will not only provide answers, but also write code on our behalf. It’s incredibly powerful, transformative technology.
But there are downsides. One major concern is that AI sometimes fabricates information. And as generative AI becomes more prolific, it’s generating massive volumes of data 24/7. Over time, machine-generated data may outnumber human data, which could distort the digital landscape. We must ensure the AI doesn’t perpetuate falsehoods it has previously generated.
Looking further ahead, this shift raises deep questions about the future of human work. If AI systems can outperform humans in many tasks without fatigue, what becomes of our role? There may be cost savings, but also the very real risk of widespread unemployment.
AI also powers the Metaverse, so progress there is tied to improvements in AI capabilities. I’m also very excited about synthetic biology, which could see huge advancements driven by AI. There’s also likely to be significant interplay between quantum computing and AI, which could bring both benefits and serious challenges.
We’ll see more Internet of Things (IoT) devices as well—but that introduces new issues around security and data protection.
It’s a time of extraordinary opportunity, but also serious risks. Some worry about artificial general intelligence becoming sentient, but I don’t see that as likely just yet. Current models lack causal reasoning. They’re still predictive tools. We would need to add something fundamentally different to reach human-level intelligence. But make no mistake—we are entering an incredibly exciting era.
Adopting new technologies can be both an opportunity and a risk for businesses. In your view, how can organisations strike the right balance between embracing digital transformation and making strategic, informed decisions about AI adoption?
I think it’s vital to adopt the latest technologies, just as it would have been important for Kodak to see the shift coming in the photography industry. Businesses that fail to even explore digital transformation risk being left behind.
However, a word of caution: it’s easy to jump in too quickly and end up with the wrong AI solution — or the wrong systems entirely — for your business. So, I would advise approaching digital transformation with careful thought. Keep your eyes open, and treat each step as a deliberate, strategic business decision.
When you decide that you’re ready to adopt AI, it’s crucial to hold your suppliers to account. Ask the hard questions. Ask detailed questions. Make sure you have someone in-house, or bring in a consultant, who knows enough to help you interrogate the technology properly.
As we all know, one of the greatest wastes of money in digital transformation happens when the right questions aren’t asked up front. Getting it wrong can be incredibly costly, so take the time to get it right.
Photo by petr sidorov on Unsplash
Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is co-located with other leading events including Intelligent Automation Conference, BlockX, Digital Transformation Week, and Cyber Security & Cloud Expo.
