This blog is part of my 38 before 38 series. I write a blog for every single day for the 38 days leading up to my 38th Birthday.
In a recent family gathering, my uncle was puzzled at the why the leading audit firm forces their interns to overwork. “It’s all computers now.”
It is not. Accounting requires a lot of work. There is an exponential increase in . It is a very very difficult to reconcile every invoice even in a small organization.
You would think with AI, accounting firms wouldn’t need interns anymore. or Accountants for that matter. You can give data to ChatGPT, and tell it to find discrepancies and be done with it. Besides the data protection issues, ChatGPT is not very good at parsing human error. But even if it was, the patterns are not always clear. The “word pattern recognition machine” is not as good at “reasoning”, no matter what the peddlers tell you. It is also not good at understanding basic human deception.
Assuming all the VC pipe dreams come true and AI becomes a d, Engineers and Accountants will not be out of a job because of LLMs. Because LLMs can’t survive.
Why Accountants?
Think of a business as a machine. You put resources in that machine, like labor, capital, raw material, etc. Lets say all those resources cost $X. The machine then gives you back a profit $Y over your original amount.
At least in the platonic ideal of a business. There are many factors that affect a business’ health(ie the way to get that $Y). Are you working off debt or extending credit(Assets/liabilities)? When do you receive the payment for your goods and/or services(cash flow)? And other considerations.
A business that is using LLMs or selling LLM services do still need to account for these factors. And since the LLM is bad at doing this job. So accountants will be doing
Now you might ask, “what if it gets better?”. There is a possibility of that. However, humans are resourceful and adaptable when it comes to deceit. And if not, they are also very competent at being incompetent. They find new ways to do that every day. That only another human can anticipate.
Why Engineers?
Open any text editing app on your laptop. Word, Notepad, or even a comment box on any website. Write 2 lines. Now move your cursor around using the arrow keys. Seems pretty simple, right?
The amount of thought and design work that went into making it work is mind-boggling. To account for every edge case requires a lot of work. However, at the beginning of implementing arrow key navigation, the engineers would have described this work simple as well. It is not easy to foresee a future when you have no past frame of reference. From your volume controls to how load times are calculated, everything required serious work to implement.
ChatGPT, Claude, etc can produce decent code snippets. Given the structure of programming languages, this is the only thing they may show exponential gains in. However it would require a lot of anticipation to maintain these technologies. Something LLMs are not capable of, because of fundamentals too boring to discus here.
And it’s not just computer and software engineers. You need electrical engineers to build the power infrastructure for an LLM, civil engineers to design efficient buildings. You need HVAC engineering to manage the heat generated by the servers. Nuclear engineers to run the power plants.
If the great robber barron fantasy of a labor free world comes to pass, it would not be labor free. It would require a lot of complex labor.
In the year 2000
Predictions are hard, especially about the future. Maybe there is a mathematical breakthrough, and everything I said here becomes irrelevant. But breakthroughs are random. Given the state we are in today, this is the likely future.
So what jobs are in danger of taken away by AI? We’ll see tomorrow.