I've always wondered how effective this would be in different industries. In some businesses, leadership exists to make very easily quantifiable (and perhaps therefore more automatable decisions) - answering questions like "Should we build a new factory?; What product lines should we develop?".

Where I work, in consulting, the senior leadership exists almost solely for the purpose of building relationships with customers and bringing in business - really hard for AI without a radical change in the sales model (but I can imagine a sort of Consulting-As-A-Service, given that we already bill by the hour).

I really suspect (and worry from an own-career perspective) that the first thing to be automated will be the number-crunchers surrounding senior management. I spend my time doing things like calculating the marginal effects of extra pay on employee productivity, which is hard largely because most companies have terrible data. Once companies start implementing systems that can effectively and automatically track complex metrics like this, around 80% of my role won't be needed any more, and instead of paying $250/hr for a guy to whack out some R code, companies can get the same service wrapped up in whatever future version of SAP they're using.

I heard a story once about how the introduction of power tools was protested by the set builders of Hollywood (and probably Broadway) - they said, "now it takes fewer people to build the same set! we'll lose jobs!"

The reality turned out that people built bigger sets on the same budget, and people with smaller budgets could now build sets.

A large part of the job of programming is automating away the job you're currently doing, so you don't have to do it again. This frees you up to do more, newer things.

I suspect both with hold true for you. What you currently do will be wrapped in an automated layer, and you'll do something similar on a higher layer previously inaccessible.

Yeah, that's pretty much what's going to happen. My employer just acquired a rival that does exactly that, so now my entire division is gradually shifting to build on their work and automate more and more.

The biggest change internally is what kind of grad degrees people get. 10-15 years ago everyone went for an MBA when they wanted to move up, now more people are getting their Stats PhDs, then coming back and applying their knowledge to problems that would have been the domain of B-Schoolers just a few years ago.

I agree, a lot of successful organizations have a lot of roles a that are people focused e.g customer/partner relations or leadership within the organization.

for your other point, i don't think there will ever be a concrete best-practice system for all companies to use that will do away with the data wrangling/customizing that currently happens. Given the amount of time that most ERP systems require to implement/improve I think bad data will be around for a long time.

but AI/learning based decisions could (hypothetically) work in a similar way to the algorithms built by quants in high frequency trading.

Your premise is incorrect. Before bankruptcy, Hostess's CEO made maybe 50-75x what a worker made, but for every C-suite exec there were thousands of line employees. Cutting workers and the wages of workers is where the money is at.

The nature of shareholders may provide huge incentives to continue to automate, all the way down to the most essential people required to run the design, production, distribution, marketing, etc.

Top company leads, board members, and advisors may suffice to run the operation and innovate, causing the CEO role to be phased out.

> sacking a thousand blue collar factory workers won't save much but fire the decision makers (or outsource them) and you'll save an absolute fortune.

That math doesn't really work; if the CEO makes $1.5 Million, and workers cost $10,000 per year (low-ball), getting rid of the CEO only pays for 150 workers. If you've got a top-heavy corporation, sure getting rid of a bunch of Chief whatevers and Vice Presidents is going to help, but realistically, workers are where the money is going.

Basically the topic Manna[1] talks about

[1]http://marshallbrain.com/manna1.htm)

The clearest way to think of AI and automation is in terms of a person, or multiple people, but it can also take the form of partial people. AI is already helping middle and upper management make better decisions.

Instead of removing the need for any of them it has instead increased their productivity. That is a partial driver for the increasing wage gap between occupations, and makes complete sense from an automation as personal productivity improvements.

The highest cost in almost every business is worker salaries.

And once the workers are all gone, the only people left to cut will middle and higher management.

Somebody, likely in an audit or an internal controls position, will be the last person to turn out the light.