Project management, writing & art — what could AI do to the world of work?
I was inspired to write this article after listening to The Daily’s recent interview of Geoffrey Hinton — the 75 year old Godfather of AI. Geoffrey recently stepped down from Google (originally scooped up by them for $44M in a talent acquisition in 2013) in order to give his honest opinion on what’s happening to the way computers think. In parallel, a university friend of mine started raving about the wonders of the speech-to-text capability of ChatGPT's new iPhone app. The latter is in no way underrated, by the way.
I myself have been a subscriber of GPT-4 since it launched and have followed developments in AI closely for the last few years. Nonetheless, I am in no way an authority. Here is simply a loosely held opinion on what AI might do to the world of work from someone with very little experience in the field. However, as a human intelligence, I at least am entitled to an opinion.
Premise:
The pyramid of job obsolescence has been turned on its head.
As AI has become more mainstream over the last decade, predictions have been made that it would take over those jobs that required ‘less skill’ first (eg. blue-collar manufacturing jobs or administrative roles) and work its way up the hierarchy of human pursuit — with ‘AI programmers’ at the apex of the pyramid just above other types of computer programmer, themselves above philosophers, mathematicians and scientists. Needless to say, these predictions were put forward by those very AI programmers.
With the giant leaps forward in neural network architectures and the computer power made accessible to them, what instead has happened is that art, computer science and writing (those formerly near the apex of the pyramid) are the fields most feeling the pinch of AI. Other creative pursuits such as poetry, prose, painting and graphic design are being dismantled or more optimistically, enhanced, by AI at a surprising rate too. Even scientists and design engineers are feeling the pressure as AI displaces traditional trial and error experimentation through modelling algorithms.
Only slightly further down the pyramid, translating, generally seen as 'white-collar' work, is also being wiped out. And market research and analysis, the central purview of consultancy firms, is now performed robustly and expediently by AI. Even creative marketing (think taglines, mission statements and product naming) has become accessible to AI, and in turn is being harnessed by non-professionals.
All the while ‘blue collar’ work continues to be surprising resilient. Here, as examples, I refer to almost all physical labour including practical skills such as plumbing, carpentry and construction, personal assistance and hospitality. While mid-level activities are also holding strong though certainly affected. These include business development, hands-on engineering, project management and HR.
Please note, I refer to work activities in this article as 'mid-level' or 'blue-collar' in relation to the social prestige they conjure, not anything else, and especially not inherent value.
What can we learn from the unexpected way AI is affecting our daily lives? And how will AI continue to affect the way we work?
Around 2016, I heard an AI expert give advice on what people should focus on to stay competitive. At the time, most people were frantically expecting to brush up on their programming skills. He suggested to do the opposite — become good at understanding and managing people. Because that’s where AI will have the most difficulty. And as we now see, it turns out he was right.
A common theme among workers who are being displaced, whether programmers, artists, scientists or translators, is that they all fit under the category of strong individual contributor. None of the positions require much people management or on-going, regular team cooperation, or for that matter, human interaction.
This contrasts with PAs, business managers, project managers, doctors, nurses, drivers, safety officers, emergency response crews and even factory line workers.
It’s reasonable to assume other areas of individual contribution will go next. Below I've made a summary of professions, that in my opinion, are within the purview of AI in the not too distant future;
- Deep market research, creative writing (though not journalism), mathematics, abstract theorising, philosophy, astronomy, engineering design, electronics, all editing (sound, music, writing, videography), logistics planning, roles involving private & institutional resource allocation, interior design, trading & asset management, accounting, musical composition and finally, website creation (this relates to computer science already discussed)
Arguably, psychology falls into this category too, although human engagement may continue as many will feel put off by sharing their deepest insecurities with an artificial, faceless entity (though the jury is out). Despite this, it's reasonable to assume that the responses and advice given by an AI therapist could be more relevant and effective than those given by a human, otherwise limited by bias, a limited knowledge base and a relatively small data-set available for live recall.
On the other hand, those activities that will continue to be unaffected for the foreseeable future will inherently require regular human interaction and cooperation as well as interaction with the physical (real) world. So, on that basis, some activities that will remain unaffected for now, include;
- Construction and factory work, warehousing, security and military roles, all roles that require any form of people management in business and government as well as dealmaking (eg. biz dev & sales), project management, management of engineering teams, HR roles, personal assistance, procurement, M&A, supply chain management, marketing campaigns and legal court-room practice.
Other safe roles will include any work that involves galvanising and inspiring people - so campaigners, politicians, leaders and visionaries. And finally high-end roles that combine expertise and real-world physical interaction eg. experimentalists, most medical professions, assembly engineers and electricians, plumbing and carpentry, large vehicle operators and offline artists & sculptors.
Now there will be a lot of AI tools that will streamline even these professions and make them more productive, perhaps displacing some of the manpower needed. But in all of them, significant human supervision will be required for the foreseeable future.
Taking a longer term view however, even those roles that today require significant real-world interaction will be at risk from AI as advances in robotics continue. These advances will remove the need for the fine motor skills, stamina, common sense reasoning and environmental awareness provided by humans. Those roles that require significant teamwork will too be eroded as distinct AIs learn to interact and work with each other as well as humans.
On the latter, which is the most exciting (advancement in robotics is a given), cooperation of AIs will embody a different type of productivity but one that is valuable nonetheless.
Let’s take the example of project management. Here it’s feasible that an AI could cooperate with other;
- Engineering design AIs
- Visual AIs
- Programming AIs
- Resource optimisation algorithms
- Prose/writing AIs
- Text to speech algorithms
- AIs for electronic design
- Modelling AIs (heat, thermal, aerodynamics, structural)
- Trial and error systems modelling algorithms
- Market research and analysis AIs
Doing so, an AI could replace said engineering project manager outright, not by learning to become an effective ‘team leader’, but by embodying the team itself through effective cooperation with a set of alternate and complementary AIs. In this universe, a person with no engineering background might ask an AI project manager to;
- Design a 1-person electric, off-road three-wheeler that is submersible, sleek, fun to drive, provides Level 3 self-driving capability, CarPlay, general connectivity, a range of 400km with 3.5s 0–100 km/h acceleration, fast charging capability with solar charging redundancy, a first aid and emergency repair kit, customisable assets such as colour, trim, bespoke paint-jobs, tyre and rim type, and retractible aerofoils, using only off-the-shelf components, over 90% of which can be robotically assembled, with a production time of less than three weeks and all for a manufacturing cost of less than €15,000 per unit.
Needless to say I may have got carried away above. But the point is, it’s feasible that an AI could produce multiple full engineering designs and parts lists for such a product within a matter of hours, rather than the months it would take a human-powered team. It is also feasible that the AI could then present these designs, as well as the component cost, assembly and manufacturing strategy for each one in the form of a narrated powerpoint presentation and written report.
How humans will work and live, and how they will interact with AIs in the distant future is something I can't predict. However, as AIs continue to integrate into even the physical world, adopting humanoid forms and facial features along with fine motor skills and common sense reasoning and environmental awareness, there will be no roles safe from AI. Certain visionary and leadership roles could remain open, but only due to our understandable reluctance to work with and follow non-human intelligences. And here, the morally complex issue of bias and ostracism against these artificial intelligences will arise. That, however, is a conversation for another, distant day and one for which we’ll undoubtedly call up Asimov’s AI to help speed things along.
