Become a fan of Slashdot on Facebook

 



Forgot your password?
typodupeerror
×
AI Technology

AI Engineers Report Burnout, Rushed Rollouts As 'Rat Race' To Stay Competitive Hits Tech Industry (cnbc.com) 36

An anonymous reader quotes a report from CNBC: Late last year, an artificial intelligence engineer at Amazon was wrapping up the work week and getting ready to spend time with some friends visiting from out of town. Then, a Slack message popped up. He suddenly had a deadline to deliver a project by 6 a.m. on Monday. There went the weekend. The AI engineer bailed on his friends, who had traveled from the East Coast to the Seattle area. Instead, he worked day and night to finish the job. But it was all for nothing. The project was ultimately "deprioritized," the engineer told CNBC. He said it was a familiar result. AI specialists, he said, commonly sprint to build new features that are often suddenly shelved in favor of a hectic pivot to another AI project.

The engineer, who requested anonymity out of fear of retaliation, said he had to write thousands of lines of code for new AI features in an environment with zero testing for mistakes. Since code can break if the required tests are postponed, the Amazon engineer recalled periods when team members would have to call one another in the middle of the night to fix aspects of the AI feature's software. AI workers at other Big Tech companies, including Google and Microsoft, told CNBC about the pressure they are similarly under to roll out tools at breakneck speeds due to the internal fear of falling behind the competition in a technology that, according to Nvidia CEO Jensen Huang, is having its "iPhone moment."

This discussion has been archived. No new comments can be posted.

AI Engineers Report Burnout, Rushed Rollouts As 'Rat Race' To Stay Competitive Hits Tech Industry

Comments Filter:
  • Why? (Score:5, Interesting)

    by Local ID10T ( 790134 ) <ID10T.L.USER@gmail.com> on Friday May 03, 2024 @04:38PM (#64445922) Homepage

    He suddenly had a deadline to deliver a project by 6 a.m. on Monday. There went the weekend. The AI engineer bailed on his friends, who had traveled from the East Coast to the Seattle area. Instead, he worked day and night to finish the job.

    Why would anyone do that? Especially an experienced "AI engineer" who should be in demand job-wise in the current market.

    Don't accept unreasonable demands from employers. It just tells them that you don't value yourself and that you will happily be mistreated further in the future.

    • Pointless to ask without knowing what compensation he's getting.

      What the deal with silicon valley jobs used to be was, you'd get a big salary, but you'd work a lot.

      • Re:Why? (Score:4, Insightful)

        by jhoegl ( 638955 ) on Friday May 03, 2024 @04:51PM (#64445994)
        That, I think, is the lie people are told.

        You get paid for 40 hours of work... YOU decide on working more, not the company.

        You get paid for your knowledge, expertise, and abilities. not because they can squeeze you for 80 hour work weeks.

        That cocaine driven mindset of the 80s and 90s is done, its dead... 40 hours is all a company gets. Emergencies are for the weekends, and if an emergency happens, its 1.5x hrs worked. So if you work 2 hours in an emergency, you get 3 hours during the work week.

        Stop with this abusive mindset from the cocaine driven 80s and 90s.
        • Re:Why? (Score:5, Interesting)

          by OrangeTide ( 124937 ) on Friday May 03, 2024 @05:19PM (#64446074) Homepage Journal

          You get paid for 40 hours of work... YOU decide on working more, not the company.

          During my first year at a tech company in the 1990's, I was told in no uncertain terms that I only get to decide if I want to leave and work someplace else. This was over taking an hour for lunch off-site but leaving at 5pm. I didn't understand that working "9-5" didn't include a lunch break. So then I started working to 6. Then I started coming it at 8. Then I would work a few hours on Saturday. Then I'd write up status reports for my team on Friday nights. And then I'd do bug scrubs on Sundays. Ooops ... my wife didn't agree with my career oriented life-style.

        • Some high tech companies still pay high performing employees extremely well, so it may be a rational decision for someone to put in a lot of extra time to get extra compensation.

          At some national labs (including where I used to work) employees put in extremely long hours without high compensation because the work was important to them. Getting called in any time of the day or night to fix a problem was common. Some of us actively enjoy our work.

          As long as people have other options, I don't see a probl
          • by jhoegl ( 638955 )
            Yeah, I dont put in work before compensation anymore. It is abused too easily.
          • Well, if they want to be paid for quantity rather than quality of work, they'll soon be replaced by AI code-bots. Working more than 35 hours a week doesn't make you any more productive, it just reduces the quality of your work & you take longer to do it. No wonder so much software is buggy & poorly designed & implemented.
        • An emergency is when you have to call law enforcement, fire service, or ambulance, & they're the ones who deal with it.

          A manager saying, "Ooh, I've just had a half-baked idea!" is not an emergency. It's an impulse. They shouldn't be able to ruin their coworkers' lives (family & social lives & health) over an impulse. Whatever the coworkers are being paid (assuming it's a liveable wage) is not part of the argument.
      • Pointless to ask without knowing what compensation he's getting.

        What the deal with silicon valley jobs used to be was, you'd get a big salary, but you'd work a lot.

        This info is roughly ten years old, but - I've known Seattle-area Amazon coders and sysadmins. Their contracts made it clear that they were expected to put in as many extra hours as needed, and that those extra hours might be very frequent.

        They were all paid extremely well, and figured (going in) they could put up with it for a few years. However most decided they'd had enough after about one year.

    • Re:Why? (Score:4, Insightful)

      by OrangeTide ( 124937 ) on Friday May 03, 2024 @05:14PM (#64446058) Homepage Journal

      Don't accept unreasonable demands from employers. It just tells them that you don't value yourself and that you will happily be mistreated further in the future.

      They all waive unvested stock under our nose. For a lot of engineering it's a way for us to retire early. Missing deadlines can tank this kind of tech startup.
      Besides, most people don't want to feel like they are letting down other people on the team. And management counts on our good nature to squeeze the most out of labor.

      • Stock options are nice, but companies are offering "AI engineers" more than $1million salaries right now. "One in the hand is worth more than two in the bush..."

        Amazon is not a startup.

        An unreasonable demand is... unreasonable. Rejecting it is not letting down other people on the team. It is respecting yourself.

      • by Anonymous Coward

        Don't accept unreasonable demands from employers. It just tells them that you don't value yourself and that you will happily be mistreated further in the future.

        They all waive unvested stock under our nose. For a lot of engineering it's a way for us to retire early.

        Retire early. Now that's a hell of an ironic goal to hear from the human engineers creating the very entity that will perpetuate more involuntary retirement than any other technical or industrial revolution in history.

        Oh the tangled Catch-22 we code.

      • Re:Why? (Score:5, Insightful)

        by alvinrod ( 889928 ) on Friday May 03, 2024 @06:52PM (#64446314)
        If you're getting an email or a call late on Friday that something needs to be done by Monday, it's your manager that has already let everyone down. Unless something is on fire, anything that's a weekend surprise is a failure of planning. I can understand crunch for a project that's been in the works for years and has an approaching deadline, but random shit that has sprung up from nowhere means you have incompetent bosses and you bailing their ass out of the fire will only teach them that they can pull the same shit again.
      • by Anonymous Coward

        Missing deadlines can tank this kind of tech startup.

        If one dude not working over the weekend will end Microsoft, Amazon, and Google all in one single go...
        Well, Good!

        Maybe you should apply for a job there and then not show up on your first day.
        You could single handedly rid the world of three out of four of the largest corporations on the planet, er, I mean startups

        You'd be a hero

  • by Joe_Dragon ( 2206452 ) on Friday May 03, 2024 @04:41PM (#64445942)

    really need unions and paid OT!

    • by will4 ( 7250692 ) on Saturday May 04, 2024 @12:34AM (#64446806)

      Companies will squeeze every hour of work time out of you, pressuring a treadmill like rise until you are working 7 days a week 12 hours a day.

      Men, young, relationship free, with certain personality types are much more willing to work those hours and increase their work hours over time.

      Women leave tech, IT, engineering, law, medical because of the ever increasing pressure to work more hours. The companies promote, encourage, leverage the agreeableness of young single men to move towards a 7 day x 24 hour always working mindset. Women leave the field because they do not want to work those hours.

      It is also why the older 35+ tech, IT, Engineering male workers are pushed out.

  • by Subgenius ( 95662 ) on Friday May 03, 2024 @04:46PM (#64445978) Homepage

    Why didn't use AI to write the code for the project?

  • by memory_register ( 6248354 ) on Friday May 03, 2024 @04:48PM (#64445982)
    There are a few useful things that AI has been shown to do. The rest is hype, and for these poor souls, burnout and frustration
    • by Tablizer ( 95088 )

      Many will merge into bigger co's. In the mad dot era around late 90's, our company's secret goal was to get bought out by a bigger co. or rival, as we knew we were skating on a sketchy business model.

    • by gweihir ( 88907 )

      While AI has some uses, they are indeed very limited. Most of what is promised as the future will not materialize from this technology and maybe never.

    • by dvice ( 6309704 )

      There are many useful things that AI has shown to do.
      - E.g. counting fish (yes that used to be a job for someone), sorting cucumbers, sorting legos... Anything where you need to detect something from a picture is perfect use case for AI.
      - Predicting weather, molecules, medical problems etc. With good enough training data, AI is extremely good with predictions, much better than humans, it has shown to be able to make predictions even from completely irrelevant data. Like diabetes from x-ray images.

      There is a

      • >AI is extremely good with predictions, much better than humans, it has shown to be able to make predictions even from completely irrelevant data. Like diabetes from x-ray images.

        Can't happen. If AI can predict diabetes from x-rays, that's not magic... there's some predictive factor in the x-rays that we haven't figured out yet. Though without looking into the claim my first thought would be it's looking at obesity as an indicator.

        Where this kind of thing gets interesting is because the AI has no actua

  • by xski ( 113281 ) on Friday May 03, 2024 @05:01PM (#64446032)
    They absolutely will.

    Learn or die.

  • There is such a crush of people who desperately want to be AI developers, that they will accept abusive treatment just to stay in that world. It's a lot like how young game developers are so enamored with the idea that they are working on *the* popular video game, that they too will put up with abuse.

    I think I'll stay put for now, in "ordinary" software development. It's not going away, any more than NoSQL made SQL go away. And with a lot of people streaming into the AI world, I'm guessing there will be mor

  • by Tablizer ( 95088 ) on Friday May 03, 2024 @06:10PM (#64446198) Journal

    Live by cutting edge, get cut by cutting edge. Tech is churn and burn.

    Warning: Semi-off-topic Rant Ahead

    Even with "regular" software, more devs and companies are more interested in chasing buzzwords than parsimony and simplicity. The result is a moving messy target. There are tools from the 90's that dev's are 4x more productive under because they are integrated tools rather than glued-on layers, requiring about 1/4 the code per feature. You don't need to worry about "separation of UI and biz logic" because BOTH are so compact that there is almost no down-side to mixing them in the same class. The separation-of-concerns movement was to manage bloated stacks & teams better, not an evolutionary step up.

    They are not web-scale and not mobile-friendly, but that turned out not to matter. Internal biz didn't need mobile after all, and we spent all that bloat and trial-and-error trying to get dual-device layouts to work right in vein. People like to tell stories about how such tools became problematic when they needed "enterprise scale", but most our internal apps are not enterprise-scale.

    The assumption is often that a web/enterprise tool can scale both down and up such that it's used on smaller projects also. Wrong Answer!* That's a failed assumption, creating bloated fragile smaller apps. One Tool Size Does NOT Fit All. Internal is not external. Desktop is not mobile. Small is not Web-scale. One of the reasons the F-35 got so expensive is that it tried to be everything to everybody. Maybe one day they'll make an affordable version, but the journey was expensive and bug-ridden.

    Humans, you are doing IT wrong.

    * It may be possible to have the small-end and big-end tools share a lot of features, conventions, and libraries, but not be the same tool. This hypothetical set should share what makes sense to share, and separate for target size when not. Do note that lack of data volume doesn't necessarily mean "simple". Billing can get rather complex for service companies, for example, but there is typically only say 30 invoiced being generated a day, including drafts.

    • by gweihir ( 88907 )

      Humans, you are doing IT wrong.

      That could not be more true. I recently looked back on the 35 years since I got my CS Master's and the overwhelming impression is that my field has utterly and completely _failed_. It is still the same old crappy mistakes being made, education is getting _worse_ and solid engineering is the exception and seems to get more rare instead of more common. I used to think "50 years until applied CS and IT is mature as a field". I do not think that anymore. Now I expect it will take another 100 to 200 years to get

      • by Tablizer ( 95088 )

        One of the reasons COBOL lives is that most COBOL apps are is not UI-tied, meaning UI fads and changes don't affect what it does. Business and administration hasn't changed much since COBOL was designed, other than the fact we normalize a little better because our hardware got faster. People aren't tempted to refashion what they can't see. Java and C# couldn't replace COBOL's lasting power without standardizing business-oriented libraries.

        • Java and other modern languages can absolutely replace any COBOL application. But why rewrite decades of code for little (if any) gain? The industries still encamped in COBOL are generally tightly controlled and financially driven. It is often a lot cheaper to support the current code for a decade rather than to switch languages.
          • by Tablizer ( 95088 )

            What I meant is that it wouldn't have the same staying power because the Java/C# biz & finance libraries would change too often.

      • by dvice ( 6309704 )

        I think we are slowly getting better. For example when I started my work, there were no unit tests. Everything was tested manually. Nowadays projects without automated tests are getting rare.
        Also UTF-8 adoption is progressing pretty well. From the top 10 million web sites, over 95% are using UTF-8. I hope that I will see the day when everyone and everything is using only UTF-8 so we don't have to worry about charset changes anymore.
        Compilers and additional tools are getting better at detecting software bugs

  • If you can't come to work on Saturday don't bother coming in on Sunday.

  • by oumuamua ( 6173784 ) on Friday May 03, 2024 @08:07PM (#64446436)
    I will take over of the job of Sisyphus himself for a $1,000,000 salary, heck some of these AI superstars are no doubt getting multiples of that, and then add in stock options.
  • Welcome to the Manhattan Project that is technology’s bleeding edge!

    We’ve all been on its thin line that cuts both ways. That’s how it got its name.

    You’re not in school anymore. Assignments aren’t neatly progressive IRL

    Ingenuity, innovation and creativity are crucible-coincidents borne of constraints

    There’s code maintenance jobs for the lackluster task-minded, park ranger or house painter

    But why only-you working when history teaches teams of three most efficient?

No spitting on the Bus! Thank you, The Mgt.

Working...