AI and the Division of Labor

AI is changing the way how work is divided in an organisation between departments and individuals creating new networks allowing different forms of division of labor. It would help if someone can give concrete examples of how this is happening? Also if someone could give explanations based on the economics of the process?

@Hardiksk21 I’ve a formal background in Economics, but must admit have not studied this particular subject with rigor.

I can offer a simple example though: Traditional ‘division’ of labor involves a high level of specialization in various roles and areas across the business.

Perhaps part of the suggestion behind the potential benefits of AI is that, by facilitating the use of ‘natural language’ prompts and inquiry it lowers the overall barriers to specialized areas of knowledge in the business.

In contrast to say even traditional search/research methods, one still has needed access to the terminology and fundamentals behind certain areas in the field (read: expertise) to even know ‘What question to ask ?’ and ‘What is the right way to ask it ?’.

A simple example I can give: Think of your typical low level tech support agent. This jobs for one, often don’t pay that well, nor do they upfront expect the employee to have years of specialized support experience. In many cases these employees are trained on or just reading from a support tree script. Accordingly the quality of the support is low and I would say, outside of additional training, the employee is really not gaining a better understanding of the underlying technology.

Whereas with an LLM agent they can repeatedly query the engine, asking for more details where they don’t understand or need more details-- Typically a task one would have to take the time to consult a colleague or manager for.

All that said, I hardly think this means knowledge/expertise is dead. 1) Despite many of these LLMs being trained on ‘all the internet’-- Well, surprise, surprise there is still a lot of specialized (i.e. think proprietary) knowledge that just does not exist on the internet, and thus is not in the train set.

  1. A lot in the popular media has made much of these systems ability to program based solely on natural language prompts. And yes, for the savvy programmer it can save a lot of time on boring, boilerplate code.

But what currently exists cannot provide deep insight or creativity. It also hardly has efficiency or pulling crucial microseconds out of the loop in mind.

Though, yes, if you’ve never written a Python program before, and you ask it to do something really simple-- I might seem amazing. A kin to magic.

In any case, this is how I see it, it lowers the overall barriers to knowledge and information transfer between roles in traditional organizations-- But, I really hope someone at the organization still knows ‘the way things are done (the hard way)’-- Otherwise in the end, you’re in trouble.

@Nevermnd Thanks for your insights. I personally believe this might be an important area of research in theoretical economics that has hitherto (unfortunately) been almost completely ignored by the mainstream discipline.

I quite well agree with the point you make. The approach that I find in corporates that are new to AI and are trying to adapt to the technology is to set up supporting teams that do the AI part and retain the experienced personnel and their hierarchy (atleast at the top and middle tiers). In fact I find a consistent debate and discussion in teh corporate world between the roles and responsibilities of the AI and business teams across profiles. This might hint at the tradeoff between specialised knowledge and the shift to multi-specialised cross-porous modes of functioning.

Thanks for your reply! Feel free to add to the discussion.

@Hardiksk21 there are professional studies being conducted around this, but these days I am more focused on the nuts and bolts that make AI work:

I will go through this @Nevermnd , My ongoing interest is in How AI will change the networks that form in the workplace, both work related and social interaction based networks, and how they lead up to the issues of division of labor. I personally think that this is one of the crucial components that makes AI function.