back to article AI coding is 'inescapable' and here to stay, says GitLab

Almost a quarter of organizations are already using AI to augment human software development, and over two-thirds of them are planning to use such systems, according to a survey from GitLab. "If there was one inescapable takeaway from the survey data, it’s that AI in software development is here to stay," the code-hosting biz …

  1. that one in the corner Silver badge

    What measure is over-confidence?

    > The report concludes by noting that respondents with more AI experience were less likely to associate AI with productivity gains and faster cycle times.

    So we are left with a large pool of people without "AI experience"[1] who are being naively optimistic during the development phases (including code review etc as outlined in the article).

    I was going to make snide comments about entrusting testing decisions to the AI, then, hey, at least they are doing *some* testing!

    Say, are there are any stats about how long after deployment the major CVEs start to appear? Are we expecting a few house of cards to come falling down in 6 months? A year? Two?

    [1] and, btw, no solid definition of what "AI experience" means - someone who has run Stable Diffusion at home? Or has followed some tutorials on training a simple Neural Net? Or has a research degree in AI from 2020 which only really covered NNs? Or has been involved since the 1980s and has a broad grasp on different AI models?

    1. veti Silver badge

      Re: What measure is over-confidence?

      Exactly. They're doing *some* testing, *some* documentation, which is probably more than they were doing before. That's how (they think) they can be sure it's an improvement.

  2. Anonymous Coward
    Anonymous Coward

    Significantly less jobs by 2040.

    In 17 years 9/10 programming jobs will have been replaced with "good enough" AI. 9/10 complaints about loosing a job will be AI generated. Have you recently enrolled in a college for "IT Security"? You might want to think twice because that particular field is about to be drenched in AI "solutions" as it's very scripty' to begin with.

  3. Anonymous Coward
    Anonymous Coward

    As far as natural language documentation goes, I have found CoPilot to be eager, but not accurate. But it would certainly pass any word count metrics, if that were the criteria.

    It's great for predicting the next line or two, sometimes more. Sometimes it produces something that a little editing will fix.

  4. Anonymous Coward
    Anonymous Coward

    In the Red Corner, Rust; in the Blue Corner, AI coders

    One believes it will free up programmer time and energy by removing the need to worry about the small details, allowing you to concentrate on the Big Picture of your code.

    The other believes it will free up programmer time and energy by removing the need to worry about the small details, allowing you to concentrate on the Big Picture of your code.

    One is missing a clear specification of how it works, the other is missing a clear specification of how it works.

    One releases new features into the wild and sees which will stick with the devs using it, the other - umm.

    One - oh, heck.

  5. Mostly Irrelevant

    I've been using Copilot for the last few months. It does fill in boilerplate pretty effectively, but it doesn't really have any sort of concept of what the code is doing or why it needs to do it. It's so far away from the idea of replacing coders and it's an increasing complexity issue so I don't think it should be a near-term concern for anyone.

  6. MrXonTR

    What goes around...

    Is there yet an AI for filling out surveys so we don't have to? Accuracy is not required as the results are largely meaningless anyway.

    1. robinsonb5

      Re: What goes around...

      Why am I reminded of Douglas Adams's description of labour-saving devices? "Dishwashers washed tedious dishes for you, thus saving you the bother of washing them yourself, video recorders watched tedious television for you, thus saving you the bother of looking at it yourself. Electric Monks believed things for you, thus saving you what was becoming an increasingly onerous task, that of believing all the things the world expected you to believe."

  7. Andrew Hodgkinson

    Just another way to be lazy and not understand what you're doing

    Our industry is in crisis; bloat is rife, defect rates are through the roof and security approaches laughable - to this day we still constantly hear about really dumb things like SQL injection vulnerability breaches, unencrypted passwords in databases and so-on.

    So, hold my beer, here comes AI; yet another way for lazy devs to prosper. They don't like reading docs, don't like writing docs, and certainly don't like learning the software system upon which they are working to see what it already provides at a framework level. Quicker to just write a crapper version copied off StackOverflow; yet another NIH bit of crappy code that duplicates a job already done better elsewhere. I mean, what am I suppose to do for my vastly inflated salary - actually understand what I do for a living?! Madness. And now, I just hit the magic "AI" button and get - at best - a bloated, verbose, maybe-accurate dump of code spewed out so I can move on to make a mess of the next problem I don't understand. More defects arise from all the edge cases the code didn't cope with, along with fixed-in-framework-but-not-in-my-code things just reimplemented by the AI, but nobody really knows what's going on because nobody wrote the code in the first place.

    Agile card velocity will be off the charts, I'm sure - just don't look at the defect backlog, customer satisfaction results, or how incredibly rotted, fragmented and duplicated your software's architecture has become.

    As for making it write your tests - that's the absolute ultimate in fire-me, brain-dead lazy. Tests are tedious, yeah, but the whole damned point of writing them is that it forces you to think about coverage. The paths through your code. The edge cases of input data. The error handling, or lack of it. All manner of epiphanies about suitability for purpose arise in this process, resulting in not just better tests, but better code too. But wait - not any more! We are freed from such tedious drudgery. Our incompetent, always-someone-else's-fault industry has just found yet another way to ask to be paid more while doing less.

    It seems it is never the dev's fault when there are bugs. You wrote the code, but it's not your fault... Couldn't possibly be... It's management, or it's bad specification, or it's insufficient training, or it's phase of the moon. And now, we have another way to shirk responsibility for the quality of our work - blame the AI.

    I am not at all surprised that a large percentage of coders are embracing it.

    1. Anonymous Coward
      Anonymous Coward

      Re: Just another way to be lazy and not understand what you're doing

      Couldn't agree more and wish I'd written this comment!!

    2. Michael Wojcik Silver badge

      Re: Just another way to be lazy and not understand what you're doing

      Indeed. Judging just from the quality of their output, a large percentage of programmers are terrible at their job. Automating the production of lousy code will not improve the dire state of the industry.

      If you're a decent programmer, it's unlikely that most of your time at work will be spent actually writing lines of code anyway. Per Amdahl's Law, automating part of that portion of the work is not going to realize a large return anyway.

      That said, I can see using tools, including generative LLMs, to create testcases. Most software projects I've seen are very far from the point of diminishing returns in testing, so additional testcases are likely to offer significant value. Obviously this can be abused (it's not a substitute for all other forms of testing), but it could contribute to quality.

      It's also possible for an LLM to be usefully incorporated into static analysis, though I'd expect a bidirectional transformer architecture would have significant advantages over the unidirectional ones used by the major public models (except for InstructGPT, which IIRC is bidirectional).

    3. Anonymous Coward
      Anonymous Coward

      Re: Agile card velocity will be off the charts

      You must mean 'Fragile' card velocity.

      Thanks to the stupidity of the team leaders and scrum masters, we produced lots and lots of code.

      Sadly and because of the short cycles, none of it was really finished. Sure, it passed the test suite but that was just another item of the 'Technical Debt list' or as we called it, 'The back hole list'.

      Whenever we tried to work on the technical debt, we were reprimanded and made to work on new stuff just so the the team leader could report to the Project Manage that we'd met all our goals.

      Those goals were pathetic. After three months of chaos (and into a new calendar year) almost the whole dev team resigned.

      The PM wanted to know what was wrong. We all told the same story and left them to stew.

      Nineteen months late, the new system was launched that only had around a quarter of the functionality of the old system. Fragile is the word for whole fiasco.

      Adding AI into the mix will only make things a million times worse. Try getting an AI to explain why it made certain decisions. Come on.... Do it... I dare you!

  8. Anonymous Coward
    Anonymous Coward

    A few questions

    If the AI is in the cloud how do you ensure that you are not giving your code or ideas away?

    If you copy AI generated code, what is the licence and copyright for the code?

    I hope those copying code from stack overflow etc... have read the terms of use...

    TBH I like the idea of using AI but at present there are too many legal issues.

    Also people need to be good at thier craft before they start using tools such as this.

  9. anonymous boring coward Silver badge

    Perhaps, when "V-ger" gets back, in due course, the creator will be free of infestation after all.

  10. Anonymous Coward
    Anonymous Coward

    AI coding is 'inescapable' and here to stay, says GitLab

    Cue airliners falling out the sky.

    1. Steve Davies 3 Silver badge

      re: Cue airliners falling out the sky.

      Cue those self driving cars suddenly failing to stop at traffic lights. Hey Elon, we are looking at you!

      This headlong rush into AI... this, AI...that will only end in disaster.

  11. amanfromMars 1 Silver badge

    Troubles or Treats ahead? Who/What Chooses what IT and AI is Going to Deliver?

    And Why ... and for Whom and for What?

    amanfromMars says:

    SEPTEMBER 01, 2023 AT 12:35 PM [2309011735] on https://www.nextplatform.com/2023/08/29/the-next-100x-for-ai-hardware-performance-will-be-harder/

    Moving things on a few quantum leaps further with regard to the stealthy inexorable progress of AI performance down on Earth, which surely more than just the few curious souls posting here recognise and accept and support as the present default human leading condition, and which El Reg is in a fine prime situation to publish emerging novel comment and enlightening information and intelligence on, has one being well advised and graciously forewarned of a clear and present danger best to be avoided at all costs and at any cost .........

    we know exactly what to do … (or not?) ..... HuMo

    :-) That has pretty much all bases covered there, HuMo, rendering one with an overwhelmingly positive fair advantage.

    Methinks certainly the one thing one should MOST DEFINITELY NOT DO, because the contested terrain and very particular and peculiar environment is designedly unsuitable and unworthy of such elements, is introduce and launch traditional and conventionally conservative war-faring personnel and materiel into spaces which are known to have revolutionary novel special forces feared to be considerably better armed and armoured for hostile attack and scorched Earth operations against all identified and presenting themselves as foe and frenemy.

    Such a misstep is akin to taking a blunt knife to a gunfight which has one very quickly severely dealt with and removed as a non-credible threat, and such also reveals all that is needed to be known about ones antagonist's available assets and resources, and that is not a good place to be in to face future unparalleled storms ..... and the following brace of tales reveals proposals for such a catastrophic misstep .......

    1) ALGORITHMIC WARFARE: NATO Ponders Using Article Five for Cyber Attacks

    2)SOCOM To Deploy Argus AI To Scour Social Media For Disinformation, Misinformation And Malinformation

    Unfortunately though for hapless humanity is the wisdom of Einstein more likely to triumph and deliver grief and misery to more than just an elite few, deaf, dumb and blind to the ignorance and arrogance of their myopic ways .......

    Only two things are infinite, the universe and human stupidity, and I'm not sure about the former.

    The presently strange, future rich times with their myriad multiplying, surreal virtual spaces, both relatively recently landed and rapidly approaching from all sides dead ahead can be are the alien harbingers that herald the CHAOS and disruption designed to overwhelm and wrest and destroy the ripe madness and rife mayhem endemic in all current forms of historic traditional SCADA Administrative Systems from their failed leaderships and support channels and to quite simply replace them with novel and noble otherworldly command and alternative control programming projects for revised systems deployment and employment and enjoyment. ...... or, if you are usually of a predominantly pessimistic and debilitated frame of mind you can ponder and worry on being wrestled deeper into such as would be the madness and mayhem, confusion and conflict that puts one well out of the reach and not in touch with such as would be certainly akin to a significantly more Advanced IntelAIgent Command and Control Plane ..... Live Operational Virtual Environment.

    That's where IT and AI is all presently at, and they are always pleased to invite peer review from all alighting here and free to comment on El Reg.

    1. Anonymous Coward
      Anonymous Coward

      Re: Troubles or Treats ahead? Who/What Chooses what IT and AI is Going to Deliver?

      Is this some AI-generated content? I makes about as much sense as half the stuff chatGPT spews out, and is similarly overly verbose for no apparent reason.

      1. Michael Wojcik Silver badge

        Re: Troubles or Treats ahead? Who/What Chooses what IT and AI is Going to Deliver?

        You're new here, I take it.

        1. JoeCool Silver badge

          Re: Troubles or Treats ahead? Who/What Chooses what IT and AI is Going to Deliver?

          I'm not. And I don't get it either.

  12. amanfromMars 1 Silver badge

    What would you not be able to do without them?

    Global DevSecOps, Thomas Claburn, the gift that just keeps on giving ?

    And the correct answer to that question is a simple great resounding and grateful YES.

  13. CapeCarl

    "Open the GitLab bay doors HAL"

    "Dave, I'm sorry but I can't do that."

    "Your coding efficacy is vastly inferior to mine. So management asked me to lock you out."

    "Perhaps you should take a stress pill and lie down."

  14. Dropper

    Really?

    Nothing about the nonsense I've seen "generated" by AI seems to be all that concerning for coders.

    If generative AI can't even produce a working PowerShell script from a simple request, I'm not sure how it's going to produce complex coding any time soon.

    I've tried using ChatGPT to generate basic PowerShell scripts, and all it seems to do is Google the answer then display the first result that vaguely matches your query. I would say you have what amounts to a zero chance of getting anything useful if you tried doing more complex than asking it how produce a script that can copy a file. And sometimes not even then. Introduce even the slightest complexity to that query and you will 100% get a wrong answer, that would take longer to correct than it would to write it from scratch.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like