The harsh reality of tech in finance: UBS calls coding an outdated skill | Hacker News Reader

The harsh reality of tech in finance: UBS calls coding an outdated skill

efinancialcareers.com

22 pointsErynPeters41 comments

This isn’t UBS saying this, it’s their “chief economist,” and you have to keep in mind that title inflation is rampant in banking (they are famous for handing out VP titles, director titles etc).

The tell that this is not a core executive is that this comment was made in a blog post the person wrote with a bunch of other spicy statements. How many investment bank shot callers or revenue producers spend their time writing blog posts?

Chief economist is basically a marketing/pr job. You’ll see them frequently on cable news, op-ed pages, etc.

I don’t work in finance btw but if you read any business news this is pretty obvious.

This. Economists have had some of the most ludicrous takes on automation Ive ever seen.

There was a ball state U paper for instance that mathematically confused offshoring with automation: https://www.nytimes.com/2016/12/21/upshot/the-long-term-jobs...

There was one about 3 years ago from HSBC where they made a "creativity score" and assigned it to each job and just assumed that "low creativity" jobs would be automated: https://finance.yahoo.com/news/these-are-the-jobs-with-the-h...

Key takeaway - just 7 more years and we will no longer need construction workers!

I dont think they actually ever dig down into the nuts and bolts of the economics of automation or try to understand it on a low level and their knowledge of engineering is childlike. I think this guy in particular is imagining a certain level of literal magic going on under the hood of GPT which definitely does not exist.

My guess is that this is all an elaborate way of saying that layoffs are coming and that shareholders should just assume that it's pure profit for them because AI is magic.

I can’t remember where i read that economists are modern day sooth sayers, doing the modern equivalent of killing birds and studying the layout of their entrails to devine the future, with a similar probability of success.

Here’s the original post: https://www.ubs.com/global/en/wealth-management/insights/chi...

It’s almost a throwaway line in a quick update.

This isn’t a UBS official statement — it’s a guy who works for UBS having thoughts on his UBS-published blog. I imagine he hasn’t actually used systems like GitHub Copilot himself, or read studies by big tech companies about how much such systems actually accomplish in terms of productivity. Current benefits are currently pretty marginal in big tech.

In VC and finance land Copilot is going to make all developers 10x more productive and any company not using it is going to fall behind. If CTOs are not buying Copilot for all their developers they are going to get fired.

There is a huge bubble brewing and the only beneficiaries are MS and nVidia. Billions spent on ventures that do not make anyone money. Dotcom bubble 2.0.

But it says 55% faster coding right there on the website: https://github.com/features/copilot

Never mind that number came from an in-house experiment with deeply artificial conditions where the study was designed to show the product in the best light.

If you can replace high-level developers and quants and "STEM jobs", you can replace the business people too. AI already does a great job of writing bullshit executive communications and docs.

nonsense. even if you were to replace middle management, surely there would need to be human leadership. some things just can’t be automated.

(this message has been approved by human leadership.)

The only reason that we can't replace the executive class with AI is that Boston Dynamics has yet to demonstrate a robot that can navigate a golf course.

Take my upvote, damn you. Made me spit out my drink.

that’s what caddy’s are for. massive dynamic solving the wrong problem, or lying.

the robot just needs to be able to follow instructions from a designated individual. a trusted advisor, if you will.

then it will all be over.

Last year or before, someone posted an article here about using AI to replace middle management or even higher up management and the general consensus was this was stupid and impossible.

But LLMs are apt at producing nonsense???

It kind of is stupid and impossible. You can't let computer make business decisions where somebody needs to be accountable (sometimes even criminally) for such decision. So I am not worried about any managers and I am not worried about accountants. If that ever become a possibility in the future, then any crook can hide behind AI.

Man who wants to pay software engineers less: “Coding is an outdated skill”

“Since FORTRAN should virtually eliminate coding and debugging…”

FORTRAN preliminary report, 1954

http://www.softwarepreservation.org/projects/FORTRAN/BackusE...

"Coding" meant something different back then, didn't it? The enumeration on p. 2 has it listed as an activity distinct from "programming".

Funnily enough, LLMs are a lot better at producing good-sounding economic analysis than at coding. Software will continue to eat the world --including finance-- and that requires good coders.

As for this UBS chief economist, maybe he's ready to replaced by an LLM:

> In the fourth quarter of 2023, UBS's investment bank made a loss of either $169m before tax, or $280m when "items that management believes are not representative of the underlying performance of the businesses' are added back in."

https://www.ubs.com/global/en/wealth-management/insights/chi...

> STEM skills have traditionally become obsolete more quickly than other skills. It is easier for technology to replace the absolute skills of STEM than other more abstract subjects.

What are those 'more abstract subjects'? If you remove STEM from civilization you are left with social engineering, power plays, sorcerers, and snake oil salesmen. We tried that centuries ago, and i don't see any reason why economists are not replaceable like every other skill. It's a universal approximator, and that is proven by hard STEM science

Electricity is coming from the wall and we can buy iPhone and Mac in Apple store, so what we need those STEM guys for? Nothing. /s

Outdated skills make huge money when big companies eventually realize they fucked up and become desperate. Fortran developers are making 200k+.

Maybe someday AI will replace developers. LLMs are neat but they aren't that great yet.

I think by the time AI replaces developers, It would have replaced anything else white collar

Software development in banking is second tier anyway. Usually people go to faang tier. If you are good enough, you could get double the pay as software engineer for props trading. You can double the pay again if you are good enough to be a trader, who needs to code btw

I don't think any aspiring software engineers want to work as the cost center in the banking sector unless they are paid accordingly

Nope. Especially if they're aware of the history of that sector and how quickly they may dump them like they did the cobol coders that they later had to hire back.

Yeah, unless you're like, in leadership or building out something new for the bank I haven't heard a lot of great things about working in the sector.

I think in that parallel universe developers would have switched to jobs akin to ghost busters. Hunting down bugs that have evolved from simple incorrect lines of code into living creatures with a mind of their own.

It's not too far off. At a recent job I realize more and more, almost half of what I was doing was tracking down defects (some in 3rd party code, not ours) rather than building new products. At another I was diagnosing and fixing issues in some distributed cloud mess almost as much as adding new features.

LLMs are not capable of replacing developers anymore than Google Translate is capable of replacing professional translators.

It's also much more dangerous to replace developers with an LLM than to replace your translator with Google Translate. Code that is close to correct often doesn't work at all. A bad translation can still be (mostly) comprehensible even if it makes your company look unprofessional and incompetent.

Although this is a bad take overall, there is some truth to the idea that tech skills become obsolete faster than skills practiced in other areas of the market, and likewise tech assets become obsolete faster than a lot of physical assets.

People construct buildings to last for a hundred years or more, but a lot of developers assume that within 10 years, most of the stuff they're building now will have to be re-built with more up-to-date technology.

But that is simply a result of a high-growth segment of the market. It doesn't mean that software engineers are any less valuable. It simply means that in order to continue to remain in the business, they need to be able to adapt to frequent changes.

I bet he configured a dashboard on some BI that somebody else configured for him and now thinks that "coding" is trivial.

I've seen this attitude a few times with accountants that learn to use Access. One of them named his PKs "useless" because he didn't know why they were needed He also joined tables with the values, not the PKs.

Another accountant I witnessed write a web system with very secure 2FA, that actually meant nothing because the whole thing was prone to SQL injection.

The title misquotes UBS's Chief Economist. Here are the only two quotes it includes from him:

>"much of computer coding already has the look of a stranded asset," something that "loses its value earlier than expected."

>Donovan calls all STEM skills "vulnerable," and notes that they become "obsolete more quickly than other skills."

This is as much commentary on the way things have always been with coding as it is anything else. If coding is an outdated skill, then it always has been according to the rationale provided.

> Code and coding are less valuable than they used to be. As banks cut costs, decommissioning legacy applications is all the rage. UBS said today that it's on track to cut 3,800 applications between 2023 and 2026. Banks are streamlining tech and focusing on automation; fewer developers will be required overall.

Yeah, big banks have been trying to decom outdated, duplicative, or abandoned codebases for a long time. These companies have lots of single-use software floating around.

Hang on your seats while Chief Economist becomes an outdated skill

This seems a much more likely outcome

This is dumb. Until we get a general AI, STEM careers are safer than most. An LLM can write an English paper, but isn't going to do chemical research for large industrial firms or run the power grid. I think if it as a tool kind of like a calculator that is incredibly useful in some fields (marketing), but far less useful in others. We still need lots of software developers too.

All of this is fine be until an outage/bug takes down a critical service, and there are s no one with the years of knowledge to fix it

That's always been my take - a monkey can run a well oiled machine - but when it goes wrong ...

Regardless of whether he's right or wrong (in fact, the actual quote and assessment he gives is right, this headline is clickbait)

But the denial and just straight inability of people in this forum to actually address the topic, and instead just offer ad hominid fallacies and snark is... depressing.

Completely expected defense mechanisms but depressing.

Rich coming from a job that is basically just marketing, and where being wrong 49% of the time is acceptably good.

Yes, it’s possible that this man is only telling us these things because he wants them to be true.

It is also possible most people vehemently opposing the view here are doing the same thing.

UBS calls driving an outdated skill.

Reply on news.ycombinator.com