Finance is better than tech at withstanding AI

Billion with a B's picture
Rank: Human | 16,302

Based on new research, I'm assuming Finance is one of the few high paying jobs with a low correlation to AI replacement given its high sales nature. Anything coding related shows high correlation with falling victim to replacement AI, because of its pretty standard ability to learn and process logical patterns.

Study

Thoughts?

Comments (24)

Nov 20, 2019

AI is overrated in general. There will always be people creating the models.

Operations across all industries is what's at risk. The product to watch out for is the "excel" of data aggregation and validation. So products like tableau and alteryx.

    • 1
Nov 20, 2019
Chunkybannanas:

There will always be people creating the models.

That's kind of the point of AI. There won't always be people creating the models.

    • 1
Nov 20, 2019

Yep. The entire premise of Machine Learning is exactly this. The algorithm can iterate itself and "learn" from each iteration. There's studies on discrete, iterative models going away and being replaced by differential models, which are much more akin to the human brain.

"The three most harmful addictions are heroin, carbohydrates, and a monthly salary." - Nassim Taleb

Most Helpful
Nov 20, 2019

AI is super super overrated. Quickbooks can't even auto make journal entries from bank statements and these tech nerds think the largest most complex transactions in the world are going to be automated...gtfo

    • 5
Nov 20, 2019

DB wants to automate back office jobs but then again they have been axing jobs throughout the summer. Interesting to see if this will actually work

DB Link

Nov 21, 2019
leonardo dicaprisun:

DB wants to automate back office jobs but then again they have been axing jobs throughout the summer. Interesting to see if this will actually work

DB Link

Surprised this hasn't been happening at a larger scale earlier across all banks. I guess it's not much of a surprise considering how bad the tech platforms or whatever you would call them are across banks.

Nov 21, 2019

Maybe this is just the experimental stage and not scaled accordingly however with the success of automating tasks I give it by 2024 until every bank begins to fully axe BO jobs.

Nov 20, 2019

I want to read the full study before actually replying- but at first glance I was thinking something seems off with certain data points. I just looked at the law section and Clerks and Court Reporters are in the safe zone. Wtf? They seem pretty damn easy to automate. Voice recognition is getting better, not hard to take voice and turn it into typing.

Not saying this discounts the study, just something I noticed right off the bat. I will read in more detail after work.

"The three most harmful addictions are heroin, carbohydrates, and a monthly salary." - Nassim Taleb

Funniest
Nov 20, 2019

What's with all the dick measuring posts between tech and finance?

    • 6
Nov 20, 2019

Vast majority of this board is undecided target school high school/freshman/sophomores. This is the only demographic that could seriously be considering two completely different careers.

    • 2
Nov 26, 2019

fuck you got me

    • 1
Nov 20, 2019

They're both good at withstanding AI because they both have a lot of human intuition.

Finance is mostly about sales as OP said.

Coding is almost by definition AI-proof because its the process by which we convert human ability to machine ability.

Yes there are exceptions like the most manual, basic coding which is already being largely replaced by smarter codes that can replicate themselves and so forth. But the higher value coders now are going to be more valuable, not less valuable, as AI grows.

AI will grow from dumb to smart over a long period of time. Jobs that require the most brains will be replaced last. Jobs that are repetitive and simple (i.e. jobs that have been called "robotic" for a decades) get replaced first

Nov 21, 2019

Regarding coding, we've seen this advancement for some time.

For loops made it possible to iterate through an array of data. Then recursion made that easier and faster. Then algorithms. Now we see discrete machine learning taking it to another level. We're beginning to see differential machine learning take off. Code advances. The difference is now it can be written to make inferences with abstract data.

"The three most harmful addictions are heroin, carbohydrates, and a monthly salary." - Nassim Taleb

    • 1
Nov 21, 2019

Addressed that in the next sentence when I said "replicate themselves and so forth." I didn't have the jargon people like to use, because I think that has the effect of overstating how powerful it all is in terms of replacing humans. Sure it has advanced to another level but I think what that does is add economic value to the human who has the intuition to design something new.

So yeah there are some more low-value coders who get replaced by that, but what I was trying to say was that a lot of that has been felt already.

Nov 21, 2019
PteroGonzalez:

Coding is almost by definition AI-proof because its the process by which we convert human ability to machine ability.

AI could write its own code a year and a half ago

Nov 21, 2019

Addressed that in the next sentence. I don't think it at all changes the fact that coding is indeed the process by which we convert human ability to machine ability. If someone is good at that first step, the fact that AI can now take it the rest of the way just adds more value to that person who writes the first step. In another scenario, if you're the guy who was only doing the repetitive low-value part today, then yes the AI's ability to write hurts you. Different situations. Don't think it makes the statement you quoted any less accurate.

Bottom line, if someone is a smart kid today thinking about a career in programming, does rise of machine learning make programming more or less attractive than if that condition wasn't in place? I think it makes it more attractive for that person.

    • 1
Nov 21, 2019

According to the report, a gaming manager (whatever that is) is less likely to be automated than a CEO.

Any gaming managers able to provide insight?

Nov 21, 2019

People have no clue what AI can/cannot do and have some Matrix style future in mind. AI can barely do very specialized tasks after a ton of tweaking and tuning to make it work. We are decades away from a real artificial intelligence. So forget the idea of sentient computers, AI will have an impact on jobs but for very specific tasks which are not creative in nature. So no, M&A will not be replaced by AI in the next 50 years for the relationship nature of it. Nor will be replaced all the sales roles in S&T. Trading is already electronic for a lot of stuff but the structured desks will always need someone to come up with the ideas for clients and to trade it. On the contrary AI is already impacting model validation and risk management to an extent, but the biggest impact wil be on the "administrative" jobs, like those who process stuff (e.g. PnL) and documents for the bank. Anyone in S&T front office will just have to learn some ai skills since banks are for instance starting to prototype some derivs pricing models based on ML, but that's it in my opinion.

For tech, well the ML hype is causing a boom in the demand for data scientits, so I would say it actually benefits someone who wants to do tech.

Bottom line: if you want to do banking do it, since IB and PE will not be replaced by AI in your working lifespan. If you want to do tech, even better: study ML and go work in the field with an inflated salary just because the role says "data scientints" or "ML engineer".

Nov 21, 2019

Regarding your last paragraph, I agree there is a lot of hype from people who don't really understand it. However, I think Google, Tesla, and Microsoft have amazing AI teams. Google makes some real breakthroughs in the field.

"The three most harmful addictions are heroin, carbohydrates, and a monthly salary." - Nassim Taleb

Nov 21, 2019

Of course they do. They have amazing engineers, mostly coming out of top phd programs like CMU, Stanford, etc. and those are the only ones actually worth the hype. But even here, we are decades away from a sentient computer (if it will ever exist).

My critique goes to all those companies that hire data scientists just to tick a box in front of upper managament but which don't have the infrastructure to use it and where the results of these teams are at least questionable. I think that if you want to do data science you have to do it well (big enough team, fast systems, vast and most importantly high quality data), and most companies who start to employ ML don't do that. They just hire a couple of DS or ML grads who will produce close to zero results.

    • 1
Nov 26, 2019
Comment
    • 1
Nov 27, 2019
Comment

"The three most harmful addictions are heroin, carbohydrates, and a monthly salary." - Nassim Taleb

    • 4