Generative AI: Areas of Impact

In ‘Engines of Engagement: A Curious Book about Generative AI’ we talked quite a bit about failure: failures in technology, for sure, but also in our conceptions of how it operates, and the potential that it holds. We wrote that book to inform the stage of debate, because it’s not always clear what landscape we are operating within.

Whilst Generative AI will impact many aspects of our lives, the impacts will not always be equal or necessarily obvious. Today i’ve sketched out twelve areas of impact under four categories:

• Things that will be made more EFFICIENT (what we do already, but better)

• Things that will be ACCELERATED (rapid development and iteration)

• Things that will EMERGE (Fracture of taxonomies and orthodoxy)

• Things that will be DISRUPTED (constraint and failure)

This is not definitive, but rather intended to illustrate the sheer breadth of disruption. Whilst normally somewhat cautious in my predictions, i do find myself veering dangerously close to the perspective that ‘GenAI will change everything’, with the caveat that it will not necessarily do it without our permission. I am not advocating for sentience, but rather for the creativity of humans. Most of what you see here is the results of the rapid proliferation of a new class of technologies that substantially dis-aggregate legacy frames of understanding. Or to put it another way, technologies that allow us to do things differently, or to do entirely new things (which may break the old ones).

Screenshot

Efficiency – helping us to do things better

Which in this case means cheaper, with greater consistency, or faster. That includes an ability to better monitor output (quality control and trends, semantic surveillance and monitoring etc). Technologies of efficiency are likely to be procured within existing frameworks – we know what they are, and how they ‘fit’. We are selling the benefits, not the concept, in general terms. Innovation may revolve around the scale that Generative AI unlocks – ability to trawl data in new ways, faster, or wider pools of data, or to calibrate between sources etc. So it may be a case of getting ‘more’ out of what we already have, or generating ‘new’ to provide insight and understanding (again, there is a semantic tinge to this – efficiency not from creating anything new, but rather translating or interpreting what we already have – at speed and scale – and hence de-skilling an aspect of leadership).

Acceleration – this is not simply about doing things more efficiently, but rather the speed with which we innovate or deploy, prototype or think. I am particularly considering this in terms of human/system interfaces, or tech/human hybrid leadership paradigms.

Part of this is through the commoditisation of dialogue (which we wrote about a lot in the chapter on ‘Learning’ in the book), but also the democratisation of technology in that more people have access to these technologies (whether we want them to or not – but please note that some people would disagree with my point here, arguing, correctly, that these technologies are only democratised at a certain level of wealth and access. They are right, but i’m arguing it from a predominant perspective of our larger Organisations, and within that context a foundation of computer ownership and connection can be taken as a baseline.)

One fascinating aspect of Acceleration is the proliferation of meta-analytic approaches, only made possible through Generative AI, but which are not without their challengers (and challenges). This happens not in the abstract, but against a background where replicability (or lack of it) is causing something of a scientific existential crisis in certain areas. Tread here with care, but best case is acceleration.

Speed is one thing but also parallel process of channels, so ‘local’ speed, but also semantic analysis of parallel channels – or summarising and synthesising technologies across hyper broad perspectives. E.g. visualising sentiments at Organisational level, or looking for predictive or patterning uses of language in channels etc. Probably a lot of snake oil and sensitivity, but also potential for those able to navigate this (or unscrupulous enough to do it, and wily enough not to get caught).

Emergence – represents a fracture of taxonomy and orthodoxy – it considers things that are newly possible, or which change foundational conceptions. This may include new categories of solution or application, as well as paradigms of operation. For example, Generative AI may fundamentally shift our mechanisms of democracy, or law enforcement, medical diagnosis or collaboration. But not just in ways that make it more efficient or ‘better’. This may be about a creative recategorisation of systems.

Arguably ‘emergence’ and ‘disruption’ are inseparable, as the resultant collapsing of legacy domains, or recategorisation of effect has the inevitable effect of abstracting legacy frames. Much as new taxonomies may invalidate older ones.

Disruption – is about constraint and failure. The impacts of this may be through the ‘pollution of the crowd’, where dominant theories occlude or blind us to new truths, or where human exceptionalism and the ‘tragedy of our certainty’ damn us. Disruption may include the full abstraction of a skill or function (like ‘legal’ or ‘HR’, or even ‘teacher’ or ‘artist’ at the extreme level). It’s worth noting that this model of disruption is only rooted in our own constructed frameworks of reality – it’s not like being hit by a comet. We fail because we fail to adapt. Or fail to even accept the need for adaptation.

Overall this type of framework indicates to me that there is a misalignment between popular narratives around Generative AI, which tend to be quite narrow and ‘known’, and the sheer breadth of potential and disruption. I would use this framework to encourage leaders to engage in the debate – to see if they can populate it, or challenge it and work out their own framework. Whatever we do, now is the time to be exploring and experimenting. In action, in motion.