@ajsadauskas @technology lol you don't need LLMs to end up in that mess ... seen it everywhere
this post was submitted on 17 Feb 2024
208 points (92.6% liked)
Technology
34388 readers
459 users here now
This is the official technology community of Lemmy.ml for all news related to creation and use of technology, and to facilitate civil, meaningful discussion around it.
Ask in DM before posting product reviews or ads. All such posts otherwise are subject to removal.
Rules:
1: All Lemmy rules apply
2: Do not post low effort posts
3: NEVER post naziped*gore stuff
4: Always post article URLs or their archived version URLs as sources, NOT screenshots. Help the blind users.
5: personal rants of Big Tech CEOs like Elon Musk are unwelcome (does not include posts about their companies affecting wide range of people)
6: no advertisement posts unless verified as legitimate and non-exploitative/non-consumerist
7: crypto related posts, unless essential, are disallowed
founded 5 years ago
MODERATORS
NGL, was totally expecting a different last paragraph. 😂
One sentence a paragraph does not make.
@zenkat @technology Totally agree.
But.
It's a surefire way to get yourself in that mess in rapid time, when you otherwise wouldn't.
@ajsadauskas @technology I agree right up to the end. I think they'll conclude they need a more powerful LLM.
Good thing by then we'll have oracle LLM. You may only use it for writing software. But we'll definitely charge you for answering questions about life the universe and everything.
That'll be all your profit this year minus the C-level bonuses please.
Average CTO: what a steal!
And the Web interface will be that of PeopleSoft!
They will conclude that they did not actually save money by replacing human developers with LLMs.
The next CTO might realize that. If there hasn't been a change in upper-level management, they'll just double down and blame the few remaining human developers for the mess.
CTO's are incapable of self-reflection.
A fellow had just been hired as the new CEO of a large high tech corporation. The CEO who was stepping down met with him privately and presented him with three numbered envelopes. "Open these if you run up against a problem you don't think you can solve," he said.
Well, things went along pretty smoothly, but six months later, sales took a downturn and he was really catching a lot of heat. About at his wit's end, he remembered the envelopes. He went to his drawer and took out the first envelope. The message read, "Blame your predecessor."
The new CEO called a press conference and tactfully laid the blame at the feet of the previous CEO. Satisfied with his comments, the press -- and Wall Street - responded positively, sales began to pick up and the problem was soon behind him.
About a year later, the company was again experiencing a slight dip in sales, combined with serious product problems. Having learned from his previous experience, the CEO quickly opened the second envelope. The message read, "Reorganize." This he did, and the company quickly rebounded.
After several consecutive profitable quarters, the company once again fell on difficult times. The CEO went to his office, closed the door and opened the third envelope.
The message said, "Prepare three envelopes."
~Stolen~ ~from~ ~reddit~
@TootSweet @ajsadauskas They'll just completely rewrite it from scratch using a newer LLM and that will be considered normal. In those 5yrs the percentage of developers who remember the idea of code having longevity will be tiny.
I dunno, humans are more than capable of doing this already.
Outsourcing is such a mixed bag. I have 2 projects outsourced to a company in India: one is magnificent and well documented and the other looks like a crack fiend wrote it. Both work, but only one is sustainable
I don’t disagree, but I’ve heard this before. Assembly devs complaining about compiled languages. C/c++ devs complaining about every newer language. Traditional devs complaining about web developers. Backend web developers complaining about blogs/cms tools. Nearly everyone complaining about electron.
And honestly I think those folks had a point. The old stuff written when the tools were simple and memory scarce were almost works of art. The quality of software development (especially with regard to optimization) has been going downhill for decades. What ever the llms do will just be part of this trend.
The use of LLMs though is more similar to outsourcing than it is to a new technology. No one is talking a out changing how we program, we're talking about changing who does the programming.
While outsourcing has had its ups and downs, I think most companies have found that skilled technical people can't really be outsourced easily/cost money everywhere. I suspect we'll see a similar thing here with LLMs because the core compentcy that makes programmers/engineers expensive is knowing what to do, not how to do it.
Greatly put, offloading to llms is nothing like people choosing for "easier" high-level languages. They are not really easier as well, imo.
Yep. This is the old school way of thinking that leads to things being shitty and not improving. "Why change if it's not broke?" Cue Uber, Google, Netflix any tech company that replaced the old guards.
Which have all descended, or are in the process of descending, into suckitude because of business issues rather than technical ones. And trying to replace programmers with LLMs is fundamentally a business issue.
They may be failing but they have replaced the industry so it's irrelevant.
-
Do you use Yahoo or AltaVista to search?
-
Do you still use taxis?
-
Do you use Blockbuster or subscribe to a standard cable package?
I'd wager you say no to all of them. So while the old may be right, it's irrelevant because they were still outperformed and no longer exist or are just not as competitive.
Again, people get hung up on the best or right way to do things when the reality is that's not how business works.
You'd lose that wager, actually—this area has at least two taxi companies but no ride shares (Uber and Lyft have very little penetration in Canada outside a few specific cities), and our household does subscribe to a standard cable TV package, although it's mostly for the benefit of my elderly mother. Those companies have not been nearly as disruptive as some people think they have.
(As for Google's search engine, I wouldn't touch it with a barge pole these days. And Yahoo and AltaVista both sucked even when they were popular—I preferred InfoSeek, back in the day.)
Anecdotally sure, but for the majority of people I'd be right. And that's what matters - at a small level you'll have outliers but if you're winning the majority of the market then you will crush your competitors. Again it's irrelevant whether your code is good or efficient or replaced by llms so long as you are winning long enough to kill your competition.
Actually, what really matters is not the quality of your code or the disruptiveness of your paradigm, or whether you can outlive the competitors that existed when you started up, but whether you can keep the money coming. The rideshares in particular will fail over time in any country with labour laws that allow drivers to unionize—if the drivers make a sane amount of money, the company's profits plummet, and investors and shareholders head for the hills. Netflix is falling apart already because the corporations with large libraries of content aren't so happy to license them anymore, and they're scrambling to make up the revenue they've lost. Google will probably survive only because its real product is the scourge of humanity known as advertising.
Again, it's all business considerations, not technical ones. Remember the dot-com boom of the 1990s, or are you not old enough? A lot of what's going on right now looks like the 2.0 (3.0? 4.0?) release of the same thing. A few of these companies will survive, but more of them will fold, and in some cases their business models will go with them.
I actually don't disagree with you and think we're on the same page. Basically, you can summarise our whole discussion as all companies are doomed to fail at end of day.
If you don't change and innovate you will fail.
If you change and innovate too much you will fail.
Finding the middle ground is rough and most companies will fail.
They'll outsource to cheaper offshore LLMs.
I think every CTO expects to inhereit technical debt, whether by overworked devs or overautomation
I haven't seen any talk of wholesale replacement of developers with LLMs in my organisation. What has happened is that these tools have been made extensively available to developers. I think right now they are basically being assessed in terms of how much they help developer productivity. Not sure about other places though, I agree with the idea that it's not really feasible to just straight up replace devs with an LLM.
Yes, they aren't there... Yet.
@ajsadauskas @technology I've been thinking for a while that I truly pity anyone who's going to have to maintain this AI-generated code.
@ajsadauskas @technology Here's some data to back it up too 😊
https://visualstudiomagazine.com/Articles/2024/01/25/copilot-research.aspx
@veronica @ajsadauskas @technology The hype around AI in software engineering seems to be that it is ‘proven’ that devs produce code quicker. it is going to be interesting to see if the corporate world values code quality over development velocity. There seems to be a pervasive belief that “move fast and break things” is how the big guys do software engineering. A few points to note:
- this idiom only applies when you fail fast, realize it, and address the problem that has been introduced.
- Break things does not mean enshittify ie create tech debt by virtue of poor code
- It really only applies if you have enough development resources to do the rework. That is to say, can afford to get it wrong often.
#AI #copilot
Not five years, but this year itself
@ajsadauskas @technology A system that can’t balance brackets and is awful at math is gonna do great 👍
@ajsadauskas @technology …and they’ll find themselves lapped by those who didn’t drink the LLM Kool-Aid.
If quality trumped speed and convenience how does McDonald's stay in business?
@BrianTheeBiscuiteer McDonald’s actually puts a burger on the bun. Not the best, but adequate for a quick bite. General LLMs put bullshit into the ether.
Yes, adequate is the word. The tipping point doesn't come when they're "good", it comes when they're "good enough".
@ajsadauskas @technology good prediction. This is basically what they always do with every overly-hyped technology.
In 5 years time? We may be having a war against "Skynet" in 5 years. The tech is advancing at a scary fast pace.
Tha Superior Tech ™™™✓✓ writes garbage code that a 15 year old can produce after years of data hawking and information stealing. Newer models in 5 years may get at a university junior level of coding.
AI bros are on more skooma than Crypto bros
I use AI to write mundan code and I always have to almost rewrite everything anyway