Everyday AI

Welcome to the latest edition of my monthly newsletter reporting on things I discover through client work and my network. There's no fixed topic, but I am mainly involved in growth projects in financial services, digital and data businesses.

I had an AI epiphany last week. Not a great thought about the future of humanity – I will leave those to Mustafa Suleyman and Yuval Noah Harari. Nor an insider insight from Silicon Valley – for those I rely on Azeem Azhar whose Exponential View newsletter comes highly recommended. Nor still an idea for dramatically transforming your business – I’m still working on those.

No, my AI epiphany was more a realisation on how accustomed we have already become to Generative AI, which seemed like a miracle only two years ago. Technology goes from the magical to the mundane remarkably quickly. So here are some of the ways I’m using AI every day often without even noticing, in ways which were barely imaginable in 2022.

Transcribe and summarise

One morning last week, I had a video call with a contact, where we both transcribed and summarised the call using our own AI assistants. In my business, this has rapidly become standard procedure using Microsoft Teams, where the capability is well integrated and available with a couple of clicks. This allows us to create more accurate, objective and shareable meeting notes than we can write ourselves as well as saving time.

Later in the week, at a conference, the roundtable discussions I was facilitating were transcribed with remarkable fidelity and I was sent the meeting summaries the next day. Stepping this up a level, when conducting a series of interviews for research, it is now easy to create succinct summaries of hours of discussions, and even ask the AI to produce suggestions on what to do with the results – create an action plan, a workshop agenda or a set of recommendations.

Transcribing and summarising text feels like a real super-power for generative AI, where it clearly outperforms humans at least in speed, and arguably accuracy and objectivity.

Enhanced search

Google now often offers an AI summary alongside search results. Of course, there are obvious concerns about accuracy and provenance of data, as well as implications for Google’s business model (let’s let them worry about that…) However, if I’m searching for a recipe for chicken soup, a quick recap of the chancellor’s Budget, or restaurant suggestions near my next meeting the stakes for accuracy are not too high. Very often the AI summary is good enough to answer my question and I don’t need to click through the links – although they’re there if I need them.

Going beyond this, using Microsoft Copilot, I can now search through my files and emails for references and information that I can’t perfectly recall. The usefulness of this is still patchy – but sometimes it can be spot on and save a lot of time. And of course, the more files and history you have, the more value there is.

Content creation

These newsletters aside, I’m not primarily in the business of content creation, but like many other professionals I produce research and write reports and emails all day long. Of course, none of us would pass off the work of AI as our own, would we? Not least because it can get facts wrong, miss key points of nuance and balance, and produce rather anodyne prose. But it is helpful to produce a first draft, getting over that blank page feeling. Even with this headstart, I’ll still spend almost as much time editing and refining as I would have done starting from scratch. But AI can help with this process too - Microsoft Copilot will coach you on the tone and style of your emails, and I have to confess has given me some pretty good tips.

I know some people who swear by using conversation with a chatbot to shape their ideas and use as a sounding board. Personally, I don’t use it like this, but I think that’s only because I’m not in the habit.

In academic circles, there is now genuine and growing concern about students using AI to complete essay assignments. My academic friend is convinced she can tell AI generated content from the real thing, but apparently both the students and the AI tools are getting better at fooling the experts, for instance by using one AI tool to produce the research, and another to humanise the prose style and make it more like your own.

Custom image generation

The specific event that triggered my epiphany was my 10-year-old turning to me while doing his homework, saying “hey look I’ve used AI to create an image to illustrate my research on acid rain”. This use case - creating high quality custom images - really did seem amazing when it was first introduced, and now it’s just a fact of life at primary school.

For most of us, this is just a bit of fun (as you may have noticed from the cartoon illustrations I use for this newsletter), but with a bit more effort, creativity and better tools, it is easy to see how the worlds of illustration, photography, and fashion are rapidly changing.

So much more

Aside from the rapid increase in capability of the language models themselves, most of these advances have been created by consumer tech giants making it easier to adopt the technology by embedding it in applications we are already using.

Even before the Generative AI revolution, AI had crept into our lives imperceptibly through social media, personalised recommendations, smart assistants and machine learning embedded in many digital experiences and business processes.

What’s next?

It is still striking how dependent the main consumer Generative AI applications are on the chat interface. Unsurprisingly language models are still more adept with text than with numbers and graphics. Although Microsoft have integrated Copilot quickly into their Office 365 suite – in most cases it defaults back to asking questions and getting results back through a chat sidebar. Integration into Excel and Powerpoint is still crude, and deep integration into the applications is going to take a little longer. However, the direction of travel is clear, and we can expect deeper integration and richer functionality over time.

The really profound implications of AI may still be to come – I don’t think we are yet seeing dramatic productivity increases, or mass redundancies - but this already feels like the smartphone revolution, in that we all suddenly have a range of new capabilities at our fingertips which are changing the way we work and navigate the world.

I’ll be interested to hear how other people are using AI in their everyday lives – I don’t have much of a sense of whether I’m ahead of the curve, somewhere in the middle or lagging behind. Tell me what I’m missing!