It
doesn’t seem quite so scary now. It hasn’t gobbled up all our Certified
Public Accountant jobs yet. It might be evolving really fast, but it
still has a long way to go.
I
use a generative A.I. assistant (I call him Hal) pretty much every day.
Hal knows a lot. He’s like the smartest college graduate that ever
happened, sitting right here next to me. He knows something about
everything. What Hal doesn’t have though is life experience. His
answers are always confident, and most of the time they’re useful, but
occasionally he’ll head off in the wrong direction and when that is pointed out
to him, he confidently makes up a reason for why the previous answer didn’t
work and why the next iteration will. For example, I spent 20 minutes
following increasingly more elaborate, and wrong, methods to select center
focus for a particular setting on my new camera. When I finally gave up
on Hal’s answers, and excuses, and started fresh with a YouTube search, I got
the correct answer in 30 seconds. Generative A.I. is useful, like a
wrench or a hammer, but it needs adult supervision.
Agentic
A.I. will be different. Agents won’t just be answering questions and
offering advice. They’ll actually be doing things. I’m expecting
accounting-related agents to be integrated into our software vendors’ products
in the next few years. The big vendors are the ones with the resources to
develop and train agents. For now though, none of us are getting replaced
anytime soon. We still need to be the humans in the loop to make sure our
brainiac assistants are offering good advice, or in the case of agents, good
results.
Earlier,
I was concerned that good A.I. agents will fill the niche currently occupied by
new accountants right out of college. Currently we set new hires to work
doing the basic repetitive things while they’re being exposed to how we do the
more complicated stuff; until they themselves get the rhythm of it and can take
over the more complicated work. What will entry level staff do if what
they’ve always done gets automated? Where will the next round of
experienced people come from if the current crop of graduates don’t have that
entry-level position to go through so they too can be experienced?
I
don’t think that’s going to be a problem. We just have to change how we
do things. In a few years, newer employees might be put to work watching
over the agents doing the entry-level stuff. They would get some benefit
from that. But we, the more experienced people, will have to adjust how
we work with the new hires, to make sure they’re getting exposed to the next
level stuff sooner. Really, the career path for younger people coming
into the profession might just get better, and maybe even faster.
But
what about when A.I. is so advanced that it can do our entire job? What
happens when it wants my job, or Ken’s? Will our clients each want to buy
their own A.I. agent to do that work, and render an opinion on their financial
statements? Would that have the same level of public trust? Or
would the world out there rather that we the CPAs continue doing what we do,
with the assistance of A.I. agents, and we provide the comfort to our clients,
and the public, that the job has been done well.
We
can’t really see all that far into the future, it won’t hold still long enough
for us to focus, but I think we’ll be okay as an industry for quite a while
yet.
No comments:
Post a Comment