There’s so much breathless hype about A.I. going on right now, and so many examples of tools, prompts and ways of using it being shared that it can be difficult to separate signal from noise. The strategist in me wanted some way of capturing and categorising all the different potential applications for businesses to provide some sort of framework of understanding. Technologist R ‘Ray’ Wang had a useful way of categorising AI a few years back, and I liked the way his spectrum of seven artificial intelligence outcomes classified AI by the job-to-be-done. Whilst this is still useful it feels as though things have moved on, particularly with the advent of Generative AI , and we needed a new way of classifying outcomes and ways of applying it. So as a first stab I came up with this:
- Information retrieval and summary: this relates to any application that helps with retrieving information from a repository or resources, summarising or categorising information, tracking defined inputs or metrics or refining outputs
- Decision support: lots in this one including anything relating to analysing information to improve decision-making, supporting research and insight generation, the provision of recommendations (courses of action, ideas, inspiration, content)
- Informed automation: going beyond simple rules-based automation to support optimisation, personalisation, notification, responses
- Knowledge building: this perhaps builds on the first three but describes applications that support knowledge sharing, broadening understanding about a topic, and even training
- Prediction: again, lots in this one, but this is forwards rather than backwards looking and is about anticipating, predictive and propensity modelling, next best action, forecasting and scenario readiness
- Content creation: by this I mean content and assets of any kind, but this is particularly significant in relation to generative AI
- Product and service provision: based on one or some of the other six classifications, the productisation of AI, developing experiences, products and services which are based on AI
I think it’s important to recognise that these different classifications are designed to be collectively exhaustive but not mutually exclusive. In other words they overlap and can be combined. Such is the myriad of potential applications that it’s probable that I’ve left some things out and so I’d welcome any feedback on gaps in this thinking. But otherwise, I hope it’s helpful in some way.
Leave a Reply