How will AGIs affect your career?

The introduction of AGI will vastly enhance the productivity of white-collar jobs on all levels. The affected employees will find employment opportunities in newly created sectors though. This is the tenth newsletter addressing the opportunities and risks of the ongoing AI revolution. For more detailed information please read my new book: "Rise of the God Machines".

In 1750, 80% of the European population worked in agriculture and the average height of men (in the UK) was 168cm. In 2020 the average height was 178cm, which implies a significantly higher average calory intake. In theory, this should mean that at least 100% of the population would have to work in agriculture nowadays to produce the necessary amount of food. As a matter of fact, it happens to be only 1%. Efficiency gains have increased the agricultural output per head by more than 100 times since 1750, eliminating most of the agricultural jobs. The descendants of 79 out of 80 families, who worked in farming in 1750, lost their traditional employment in agriculture due to new technologies coming in place. No need to complain: these descendants usually hold better paid jobs in other industries. If we had a time machine and in 1750 would have told the elders of a village with 80 families that only the descendants of one single family would have an agricultural job in the future, an uproar would have been the result, damning the dangerous technologies.

We are facing the same situation now in terms of information processing. During my career, I was an investment banker, ran a Venture Capital fund, wrote a couple of books and was the CFO of several companies. All these jobs fall into the realm of information processing and sooner or later they can largely be taken over by an AGI. The only past professional activity of mine that could not be performed better by a computer was a student job requiring me to hang washing powder samples on doorknobs.

And even though literally everything that I have done for a living could eventually be performed better by an AGI, I am not concerned at all for my professional future. As mentioned, literally all the 79 descendants who did not get an agricultural job are now gainfully employed in the industry or in services usually earning much better hourly wages than agricultural labourers do. The same has happened to the industrial workforce. In 1950, about 48% of the UK workforce had an industrial job, in 2020 it was only 9%, a drop of 39% out of 48%. And again, these 39% and their descendants found jobs in other sectors, mostly services.

The repeating pattern is obvious: efficiency gains in one sector reduce the demand for workforce in this sector and their respective salaries. The affected people (or their descendants) learn new skills and take over a new, usually better paid role in a different sector. At the begin of the process, it is usually not clear, which new opportunities will arise. A 1950s steel worker in the UK would not have expected his grandchild to write code for a start-up or deliver sushi on an e-bike.

So let us take the angst away. We will not face an unemployment rate of 20% plus due to the introduction of AI and AGI, assuming that enough training will be available and that affected employees will utilise this offer. More interesting is the question, which areas will be affected when AGIs become omnipresent. Sam Altman defined an AGI as “the equivalent of a median human that you could hire as a co-worker“. Once this “median human” is around, what does it mean for you, what is the impact of AGIs on various job groups?

1) Call center, customer relations, tele sales and back office (80%+ reduction): OMG, you are in for a tough ride. Since this is a big cost driver for many companies and as the activity is strictly rule based, companies have been automating these functions for a long time. AGI is not the begin of automation in this field, but it could possibly mark the end point. Better start exploring new career paths now. Essentially all the work that requires access to one single computer system can be done by an LLM either now or in the very near future. Once multiple systems must be operated, it gets a bit trickier, but an RPA based solution incorporating LLMs and soon AGIs will not take too long to get implemented. The only limitation is language, as humans are better in understanding poor pronunciation and difficult accents. Also, not all languages will be available for AGIs immediately. Finally, a VIP treatment could be fulfilled by humans for premium products and customer groups.

2) Blue-collar workers (<20% reduction): Since blue-collar workers perform physical tasks and since AGIs have no physical presence, one could think that blue-collar workers would feel no impact. But the production and handling processes developed by the white-collar engineers will be improved once AGIs optimize them and this will lead to a certain reduction in headcount on the blue-collar side. Further optimised production processes will require less workers during a shift; a more efficient logistics setups will require less drivers and storage workers.

3)?General Administrative Employees (60% reduction): As soon as an AGI is available and gets access to your internal systems, it will be able to take over most tasks. AGIs will be able to create summaries, make proposals, analyse statistics, create presentations, take calls, and provide information from files. There are important limitations however, which will require human support going forward for a significant period, relating to more complex negotiations and physical activities.

Complex negotiations require a thorough understanding of the motivations of the individual participants, which makes access to personal information necessary (X does not like Y, wants to take over other role, goes through difficult divorce…), that AGIs are unlikely to get access to (at least not in a legal fashion).

Handling of physical objects is an even more restrictive condition. As far as general administration topics are concerned, this mostly concerns printed out documents, which are not available in a digitised format. Since external interactions are often performed through paper documents, this limits the use of AGIs. In practice, AGIs will probably handle the main load of analysis, but a certain human component will be required for as long as printed documents need to be reviewed. Given the strong economic incentive to digitise, paper documents are likely to disappear in the midterm future.

4) Lower management (80%): A lot of the activities of junior management center around the analysis of work performed by general administration and by providing summaries and recommendations. This happens to be the sweet spot of LLMs today and will certainly be a strength of any AGI going forward. As a result, AGI will eliminate classical staff functions almost completely. The main activity left for junior management will lie in the supervision of the few general administrative employees left.

5) Experts (40-80%): What will happen to the engineers constructing new aircraft engines, the loss adjusters assessing insurance claims, the city planners drafting a new metropolitan layout, the biological researchers finding new drugs, and all the other usually academically educated highly qualified experts? The impact on them will occur later and be probably somewhat less severe. On one hand, they often must deal with physical assets, examine people, construction sites and factory layouts. On the other hand, their skillset is usually very deep and exceeds the written documentation available on the internet, the main source of LLM wisdom. In most cases, AGIs will be used as tools enhancing productivity thereby generating 60% efficiency gains, which will result in 40% headcount reductions. There will be some cases where they will gradually take over almost all expert roles, but the migration process will likely be slow.

6) Middle and senior management (20-40%): analysis, the main strength of AGIs, forms only a small portion of the work of middle and senior management. When it comes to hiring, incentivising, promoting and firing staff, setting direction, leading negotiations and exerting political skills, AGIs have a long way to go. Especially on the leadership side, it is hard to see, how humans will relate to silicon leaders (more in a future newsletter). Overall, the direct impact of AGIs on middle and senior management will be low. Managers on that level will use AGIs as trusted advisors, doing the work performed by more junior staffers today, but will not be replaced themselves. The indirect impact will be more severe, caused by the reduction in general administrative and junior management headcount. A lower number of junior managers requires fewer senior managers to lead them. This effect will be less pronounced in production companies compared to service companies, as the number of blue-collar production workers will decline less.

While the number of employees required in existing sectors will decline significantly, it will not structurally enhance unemployment. One driver of this, as we will see in a future newsletter, is the massive needs for new jobs that will arise in other business fields. The second major factor is a demographic one: the baby boomers are leaving the work force in masses and the Gen-Z new entrants are far less numerous. This actually causes massive labor shortage problems in most OECD countries already. AGI will be a blessing to deal with this otherwise very severe problem.

Summary:

If you are a blue-collar worker, you can read this book with detached curiosity. The impact of the AGI introduction on your workplace will be delayed and limited. If you are an administrator without a college degree, you need to be worried, since more than half of your colleagues (and possibly you) will lose their job. The same applies to junior managers. This drastic headcount reduction will also affect middle management, albeit to a lesser extent, as there are simply less employees to manage. The overall number of unemployed will not rise due to demographic changes and new business fields.

Further Information:

The newsletter is an abbreviated version of the original article from my recently published book: "Rise of the God Machines: ChatPT, AGI, and the End of Human Supremacy" (ISBN: 9694792401, LIMIT 2024). If you want to read the whole book now without waiting for the future newsletters, please go to Amazon and selected other bookstores.

Alternatively you can also download it here: PDF-Version

#superintelligence #AGI #pdoom #singularity #ai #startups #humanity2.0 #vc #alignment #chatgpt #anthropic #google #gemini #callcenter #management #oecd #unemployment