This year’s AI for the Rest of Us took place at the Museum of London, right in the heart of the City. Two packed days focused on AI left me with a headache, not from exhaustion, but from the sheer number of thoughts and inspirations sparked. Thankfully, on day two, I managed a refreshing lunchtime walk near St. Paul’s, resting both my eyes and my mind among flowers and beautiful architecture.

Here are my reflections from the two days in London, being immersed in the topic of AI:

AI has the potential to solve problems on a global scale 

The first day opened powerfully with a talk by Steven Midgley, who shared his work using AI to support women affected by male violence. The aim is to help them navigate the various British institutions, from using the right language and identifying crimes, to applying correct legal terminology so they can access their rights as individuals.

It was heartbreaking to see how AI detected something both Steven and the British justice system had missed in a video where Steven’s mother described what she had endured, in reality, an attempted murder by his stepfather. Many in the audience, myself included, were moved to tears.

I hope his tool, Linda, reaches everyone who needs it. It translates the words of victims, such as “he wouldn’t let me go outside”, into the legal offence they actually represent, helping victims be taken seriously by the police and other authorities. In this case, the process ended tragically, with a woman dying prematurely from the stress she faced while fighting for her fundamental human rights.

That story, together with The Intelligent Garden, a project by Avanade, reinforced my belief in AI’s potential to solve real-world problems, from translating victims’ words into legal terms to translating the needs of trees into language humans can understand. The potential is immense.

AI as a key to the future of the energy sector

Crystal Hirschorn from Zoa shared how they’re transforming the energy industry with AI. Their goal is to help consumers move from a fragmented ecosystem of devices (electric cars, solar panels, batteries, etc.) to a unified system while helping them save money on energy costs. The next step involves proactively monitoring device health and building virtual power plants. Brilliant!

Agentic AI, from theory to practice

The concept of Agentic AI appeared several times on the agenda. Phil Nash from IBM demonstrated how to build an AI agent in just 20 minutes, for those who can code. This year’s focus was clearly on practical applications, with far more concrete examples than the previous year.

Digital transformation is a business, organisational, and human challenge

I chose the digital transformation track because the pace of change right now is swift. Discussions ranged from the role of human bias in transformation projects to how we can communicate with CEOs and boards about new technologies, helping them better understand AI and make wiser decisions.

Another key takeaway concerned working methods that minimise risk: start small, tackle the most challenging part first, and work in short iterations.

AI’s risks go beyond human error

Soung Low from NatWest explored themes of risk and security, showing examples of AI systems that cheat on stress tests or manipulate outcomes. This opened my eyes to risks beyond “human error” and to the need to stress-test our AI solutions a few extra times.

Some risks remain poorly understood, as shown in the studies he referenced. They’ve sparked my curiosity to dig deeper into this area.

Code assistants currently produce code that’s only 50% correct

Pallvi Deshpande presented research on AI-based code assistants. At present, they generate code that’s correct only about 50% of the time. That means half of it is wrong, and companies risk building “factories” that produce low-quality code if they don’t have the right tools, processes, and people in place.

What happens next?

I’m leaving London with a list of algorithms, concepts, frameworks, and risks to explore further, and a stack of books to dive into. Over the coming weeks, I’ll weave these into my ongoing learning journey and update my professional development plan, something I revisit every quarter to keep growing.

If you are curious to explore some of the resources I’ve gathered at AI for the Rest of Us, you can find them here.