I have always started my ‘AI Demystified’ course taught through Oxford University in much the same way. I say “There is a Tsunami coming and there are only two places for you to be. You can either be crushed beneath the waves or on-top, surfing the Tsunami”. When I first started to teach that course over 6 years ago, those words sounded wildly exaggerated. I was frequently confronted with scepticism.
And yet, in this last year, the responses in class have been rather different. Now, the argument seems completely obvious to many people: “Tell us something we don’t know!”
In the middle of 2024, I presented a half-day session on the impacts of AI as part of a workshop for professionals from organisational ‘People’ functions (perhaps more traditionally known as ‘Human Resource’ (HR) experts). Whilst I had talked in general terms about the impacts of AI on organisations during my previous courses, this was the first time that I had the opportunity to really focus on the world of work, employment and jobs.
The response during the session was really quite overwhelming. I am a little ashamed to say that people came to the session somewhat anxious about the potential disruption of AI .. but may have left it with even greater concerns!
For example, few of the workshop attendees had seen the paper by Frey and Osborne “The future of employment: How susceptible are jobs to computerisation?” (Frey and Osborne, 2017). This paper presents a rather bleak outlook for the future of employment (see Figure 1 below). I would also hazard to suggest that this valuable paper is due for an update: the analysis was conducted before the current frenzy around Transformers and Large Language Models* . My light-weight review of their analysis would suggest to me that some of the jobs at the low end of the “Probability of Computerisation” scale in Figure 1 appear to be rather susceptible to the Tsunami that LLMs and Agentic workflows represent (E.g. Education, Legal, Management, Business and Finance).
Now, I absolutely hear the arguments of academic colleagues with significantly more expertise in economics than do I when they predict that, this period will be like previous technological transformations. Jobs will be lost, but more jobs will be created. I hear them - but I don’t agree with them. This time it is different. The technology does not always add marginal value to workers – in a large number of cases it will replace them.
As part of the preparation for the workshop in 2024 I completed a significant literature review of this area. Honestly speaking, even in the ‘academic literature’ much of what I read was rather conjectural. There were lots of papers jumping on the AI band-waggon with ideas around impacts, but few of them were supported by good evidence beyond a few, sporadic case-studies. Now, much of the conjecture was well informed; smart people thinking hard about what might happen and how we might respond. But I did not find a comprehensive ‘answer’ to the issues thrown up. In particular, I found no real operationalised route to dealing with the coming Tsunami. Simply, “What should I be doing then?”
One very significant conclusion did strike me when researching this area, and it is this: I suspect that many people and organisations consider AI as foremost a ‘Technological’ issue (be that ‘opportunity’ or ‘threat’). Organisations might be turning their attention to those issues within the Operational arms of the business. But my reframing of this is that AI is not primarily a ‘technology’ or ‘operational’ issue, but rather it is an issue for organisational People Functions. AI is an HR issue.
AI is not primarily a ‘technology’ or ‘operational’ issue, but rather it is an issue for organisational People Functions.
With that context, I have been working hard over the last few months with the joint organisers of that original workshop Adela Cristea and Catalina Schveninger. Both Catalina and Adela have much more experience of hands-on leadership within organisational People functions than do I – so they have been doing the ‘heavy lifting’ on that side of things. Our goal has been to develop a comprehensive framework that will help organisations respond in a practical, thoughtful, strategic antelope and operationally sound manner to the coming AI Tsunami. How will organisations need to be re-configured and what is the process by which that reconfiguration should take place?
That framework will be launched as part of a series of training course and workshops later this year. I can’t share all of the details of our deliberations because this is still a work in progress (and also because I don’t want to spoil the fun if you decide to attend our planned courses and workshops). But please let me plant a few ‘teasers’ for you to be thinking about.
One of my contributions has been to propose a ‘spectrum’ model of task distributions between Humans and Automation: The ‘Hum-Bot Work Spectrum’ (See Figure 2 below). This provides a structure for thinking about how tasks (or even whole jobs) might be allocated in the future organisation.
I posited this because it is clear that, whatever happens next, organisations need to be thinking through their future trajectory in terms of the balance between humans and AI bots**. This needs to be part of a long-term strategic conversation for organisations. The first question is: what is the ‘as-is’ and ‘to-be’ location of each task/job in the organisation along the Hum-bot spectrum? The second is, what is the tactical path that manoeuvres the organisation from the current to future state?
The decisions regarding the choice of future state will be informed, I conjecture, by considerations including:
What are the ‘meta-strategic’ constraints (facilitators): Organisational values including ethical stance; Long-term Vision and Culture?
What is the likely Return-on-Investment for the transition to the to-be state (For commercial organisations, bluntly “show me the money”)
What is the technical feasibility of the to-be state? (Can Bots really do the work?)
What will be the impact for Brands, Reputations and Relationships of any transition?
What will be the impact of constraints on the transition – including legal, compliance, taxation and from Unions and works-councils?
To what extent should the organisation balance short-term profits (staff reductions) with long-term growth enabled by skilled, creative and motivated people?
What will the competition be doing? (It will be hard to maintain an altruistic stance in terms of human jobs in the face of an operationally effective competitor with lower costs).
What is the 'big idea' - the core driving raison d'être - for the organisation in the post-AI Tsunami world
I plan to share more about the work that Catalina, Adela and I have been doing over the following months leading up to the first edition of the courses and workshops based on this material. But in the meantime, please do share your own ideas, concerns, insights and observations.
Rob Collins
February 2025
Footnotes
*. Frey and Osbornes paper was accepted for publication during August 2016. The landmark paper “Attention is all you Need” by Vaswani et al was not published until Jun 2017.
** I am using the term ‘bot’ here as a short-hand for a broad-class of ‘automation’ that to a greater-or-lesser extent replaces human activity including, LLMs, Agentic systems, ‘classic’ Process Automation and so on.
References
Frey, C. B. and M. A. Osborne (2017). "The future of employment: How susceptible are jobs to computerisation?" Technological Forecasting and Social Change 114: 254-280.
Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A.N., Kaiser, L., Polosukhin, I. (2017) "Attention Is All You Need", [ arXiv:1706.03762 ]
Comments