Please create an account to participate in the Slashdot moderation system

 



Forgot your password?
typodupeerror
×
AI Programming IT

CBS Explores Whether AI Will Eliminate Jobs -- Especially For Coders (cbsnews.com) 159

"All right, we're going to begin this hour with a question on many people's minds these days, amid all these major developments in the field of artificial intelligence. And that question is this: How long until the machines replace us, take our jobs?"

That's the beginning of a segment broadcast on CBS's morning-television news show (with the headline, "Will artificial intelligence erase jobs?") Some excerpts:


"As artificial intelligence gets better.... job security is only supposed to get worse. And in reports like this one, of the top jobs our AI overlords plan to kill, coding or computing programming is often on the list. So with the indulgence of Sam Zonka, a coder and instructor at the General Assembly coding school in New York, I decided to test the idea of an imminent AI takeover -- by seeing if the software could code for someone who knows as little about computers as me -- eliminating the need to hire someone like him."

Gayle King: "So all this gobbledy-gook on the screen. That's what people who sit in these classrooms learn?"

"And I for one was prepared to be amazed. But take a look at the results. About as basic as a basic web site can be."

King: What do you think? You're the professional.
Zonka: Ehh.

[Microsoft CEO Satya Nadella also spoke to CBS right before the launch of its OpenAI-powered Bing search engine, arguing that AI will create more satisfaction in current jobs as well as more net new jobs -- and even helping the economy across the board. "My biggest worry," Nadella says, "is we need some new technology that starts driving real productivity. It's time for some real innovation.]

King: Do you think it'll drive up wages?
Nadella: I do believe it will drive up wages, because productivity and wages are related.


At the end of the report, King tells his co-anchors "In the long term, the research suggests Nadella is correct. In the long term, more jobs, more money. It's in the short-term that all the pain happens."

The report also features an interview with MIT economist David Autor, saying he believes the rise of AI "does indeed mean millions of jobs are going to change in our lifetime. And what's scary is we're just not sure how.... He points out, for example, that more than 60% of the types of jobs people are doing today didn't even exist in the 1940s -- while many of the jobs that did exist have been replaced."

There was also a quote from Meredith Whittaker (co-founder of the AI Now Institute and former FTC advisor), who notes that AI systems "don't replace human labor. They just require different forms of labor to sort of babysit them to train them, to make sure they're working well. Whose work will be degraded and whose house in the Hamptons will get another wing? I think that's the fundamental question when we look at these technologies and ask questions about work."

Later King tells his co-anchors that Whittaker's suggestion was for workers to organize to try to shape how AI system are implemented in their workplace.

But at an open house for the General Assembly code camp, coder Zonka says on a scale of 1 to 10, his worry about AI was only a 2. "The problem is that I'm not entirely sure if the AI that would replace me is 10 years from now, 20 years from now, or 5 years from now."

So after speaking to all the experts, King synthesized what he'd learned. "Don't necessarily panic. You see these lists of all the jobs that are going to be eliminated. We're not very good at making those predictions. Things happen in different ways than we expect. And you could actually find an opportunity to make more money, if you figure out how you can complement the machine as opposed to getting replaced by the machine."
This discussion has been archived. No new comments can be posted.

CBS Explores Whether AI Will Eliminate Jobs -- Especially For Coders

Comments Filter:
  • by rsilvergun ( 571051 ) on Saturday February 18, 2023 @06:36PM (#63304499)
    Since the late '70s. I mean if you're going to lie can you tell a lie it's at least plausible?
    • by RightwingNutjob ( 1302813 ) on Saturday February 18, 2023 @06:58PM (#63304545)

      That's half true. The part that's true is measured in dollars. The part that's false is measured in standard of living.

      Joe Sixpack makes about the same as he did in the 80s but he's got a lot more nicer stuff in his house.

      The reason for both being true at the same time is deindustrialization coupled with outsourcing.

      • by jythie ( 914043 ) on Saturday February 18, 2023 @07:08PM (#63304567)
        It is also a difference of GDP vs wealth distribution. When people claim that wages go up with productivity, what they are really talking about is GDP going up with productivity, which means on average pay goes up, but that average includes the top 1% who skew things rather heavily.
        • Building on what you wrote, the bottom line is that our current socio-economic-political system is distributing spending power for most people via an income-through-jobs link. There is also the 1% who gain most income from things like ownership of appreciating company stock, rent from real estate, and interest from bonds.

          True, some of the 1% are doing related intellectual labor of preserving wealth inequality against the 99% (or more likely hiring some of the 99% to do it for them) including by lobbying con

      • The house that he rents and cannot afford to buy?

        • by ranton ( 36917 )

          The house that he rents and cannot afford to buy?

          Home ownership percentage was 63% in 1965, and is 66% today. So it isn't like we have seen a drop in home ownership rates over time.

      • by peragrin ( 659227 ) on Saturday February 18, 2023 @07:21PM (#63304601)

        Miss 1980's my father earning $55k a year could afford a 2500 sqft house on five acres. Two cars (well used and not fancy) a motorcycle, 2 boats (one bought cheap and the other new). A stay at home wife and three kids.

        Me earning 50k a year 30 years later couldn't afford a house let alone a family.

        My wife and I are struggling on our combined 140k a year income. Between the house, two cars and $25,000 a year in daycare so we can work it is a struggle. The government $2000 per kid is a joke. No wonder why parents are have less and less kids. They can't afford them and work as society demands.

        • by ranton ( 36917 )

          Miss 1980's my father earning $55k a year could afford a 2500 sqft house on five acres. Two cars (well used and not fancy) a motorcycle, 2 boats (one bought cheap and the other new). A stay at home wife and three kids.

          $55k a year in 1980 is the same as $210k per year in 2023. He was making 4.4x average wages in 1980, which would be $265k in 2021. If you make $210-$265k on one income today, you can afford a 2500 sqft house outside of major cities, two cars, a motorcycle, a boat, and three kids.

          If you and your wife are only on average making 15% more than average wages in 2023, it isn't surprising you cannot afford as much as your father who was doing much better career-wise 40 years ago.

      • Re: (Score:2, Interesting)

        Joe Sixpack makes about the same as he did in the 80s but he's got a lot more nicer stuff in his house.

        Exactly. The problem with rsilvergun is him, and people like him (basically all communo-socialist types) is they measure wealth purely in dollars. The problem is, money doesn't define wealth. If you try to use money to measure wealth, you'll get a wildly inaccurate picture of the overall welfare of each individual. And by that I mean, because people like him assume that money equals wealth, they perpetually think that we're in the worst period ever when nothing could be further from the truth.

        • by Powercntrl ( 458442 ) on Saturday February 18, 2023 @10:33PM (#63304899) Homepage

          I don't know what metric you're using, but owning fancier consumer electronics certainly doesn't qualify as wealth in my book. All of that stuff is ultimately destined to be e-waste, given enough time.

          Wealth consists of things you can pass down to your children. Things with intrinsic lasting value, such as real estate, stocks in well-established businesses, and precious metals. By that measure, I'm certainly poorer than my parents.

          • by q_e_t ( 5104099 )
            So if you don't have children, you can't have wealth?
        • by sjames ( 1099 ) on Sunday February 19, 2023 @12:50AM (#63305061) Homepage Journal

          Yes, bread and circuses are cheap and cheerful. Things like healthcare, housing, and education are somewhere between damned expensive and unattainable.

          • We're victims of this problem

            https://en.wikipedia.org/wiki/... [wikipedia.org]

            Items where productivity cannot or have not been improved - 'healthcare, housing, and education' - become relatively more expensive compared to the costs of other items. Of course it's not the only factor, but it explains a significant proportion of our frustrations.

            • by sjames ( 1099 )

              To some extent, but there HAVE been significant productivity gains in all of the big sore points. It's just that the benefit of the gains are being hoarded at the top or simply squandered.

              In healthcare, lab tests are now automated and fast, imaging is digital and fast. Many formerly expensive tools that needed to be autoclaved between uses are now cheap disposables. Pharmaceuticals SHOULD see improvements due to mass manufacturing automation, computerized tracking and logistics, etc. Unfortunately, mush of

        • by Iamthecheese ( 1264298 ) on Sunday February 19, 2023 @12:57AM (#63305067)
          I'm a technophile and gamer, and I would gladly give up my phone and PC and only ever watch TV on a 12 inch black and white in return for being able to support a family on my wage. Talk about money vs wealth all you like but the damn fact is I'm poorer in real terms than someone in the 1970s who was doing my job. Food, transportation, and health care matter, and only one of those has improved in real terms for the past few decades. the fact that I couldn't afford a supercomputer like the one I'm typing on 15 years ago is irrelevant because this computer isn't bringing me things I can use and enjoy to the extent you imagine.
        • Oh yeah dude when I walk into a ratty apartment shared by three 30 year olds and they have a projector, bluetooth speaker, smartphone, and a Nintendo Switch. I'm momentarily disoriented and must silently convince myself I'm in the right place and haven't wandered into the Playboy Mansion.

      • by rsilvergun ( 571051 ) on Saturday February 18, 2023 @08:20PM (#63304715)
        unless you count his TV and maybe his cell phone. The rest of his stuff is actually a bit worse, not as well built and the parts are harder to replace. Computers are cheaper, but Joe Six pack could care less about that.

        Meanwhile if Mr Sixpack is under 45 he probably doesn't own a home, he rents. And exorbitant rates. And no, he can't move. He lives where the jobs are. Plus his healthcare and transportation costs are through the roof. And he doesn't know what to do about the kid's college. The factory's hiring, but new employees make 1/3 what he does.

        And we didn't deindustrialize. Except for China we produce more than any other country. We automated and improved productivity. And the fruits of that didn't make it to Joe Six Pack and his family.
        • by q_e_t ( 5104099 )
          There are some products, notably those involving CNC techniques, that have got cheaper and better. However, my best set of pliers were made in 1941 in Newark, NJ, and I'd have to pay a lot now to get a pair as good. Not that I bought them new in 1941! I inherited them. That counts as wealth, I suppose.
          • my best set of pliers were made in 1941 in Newark, NJ, and I'd have to pay a lot now to get a pair as good

            Channel Lock is, surprisingly, still great. It was the only brand available when I bought my biggest (4"? 5"? I forget) channel locks, but they make lots of other kinds of pliers too. Klein is also great. Neither is expensive.

            Not that I bought them new in 1941! I inherited them. That counts as wealth, I suppose.

            My tool set is somewhere around a quarter yard/garage sale stuff. Some of that stuff is weird and has saved the day more than once.

        • by ranton ( 36917 )

          The rest of his stuff is actually a bit worse

          This simply isn't true. Go to an estate sale and see how cheap products were made even 50 years ago. Even the nice hardwood furniture often needs a craftsman touch to be fully functional again. The belief that everything was made better 50 years ago is just a myth. Some things were made very well, but the same is true today.

          I am working on a 15 year old Ikea desk right now and am in a room with 4 15 year old Ikea bookcases. Each have made it through 3 moves over than time as well. These will last me another

      • It's a strawman.

        Joe Sixpack is surfing the same heap of thr tech wave of his time, then and now. Thei didn't have plasma TVs and tablets back yhen, but they had video recorders and cable.

        As for the rest: affordable housing, access to tinely and comprehensive health care, education, enough free time to have a hobby ot two... these are the crucial factors in the standard of living debate, not gadgets. And they've been declining since the 80s, too. A lot.

    • by ArmoredDragon ( 3450605 ) on Saturday February 18, 2023 @07:03PM (#63304555)

      Depends on how you measure productivity, but I tend to think that if productivity wasn't always rising faster than wages, things would be much worse than they are for most people. In other words, think about it this way:

      You use a computer to visit this website. In the 70s, we didn't even have computers that would be powerful enough to even display this website. Yet they cost a hell of a lot more money both in terms of their literal price tag at the time, and even adjusted for inflation. It wasn't unheard of for businesses and even individuals at the time to spend $5,000 on a computer that isn't even remotely powerful enough to display this site. Yet who spends $5,000 on a computer anymore? Basically nobody.

      And do you know how this happens? Because the tools we use allow us to make better use of our time, including the computers themselves. So we can become more productive with the same or even less knowledge than we required in the past, and while spending less time. However, if we billed based entirely on how productive we were, then instead of those computers coming down in price, they'd just continue going up in price. And then fewer people would even be able to afford them.

      Even your idol Karl Marx realized this, though in his own broken way. Basically his idea was what he called the labor theory of value, i.e. basically how much time you put into your work should determine its value. Though the broken part is that he's also wrong in the sense that not everything you spend a lot of time on is going to be equally as valuable as something you might have spent a lot less time on. Or that a lot of labor involves trial and error, where the time spent on errors can be and often is totally wasteful (the only value you gain from errors is in the form of knowledge, and even then it's not always useful knowledge.)

      People like you don't understand a word of this though, and that's why you advocate for socialism, which 100% of the time it has been tried resulted in a more destitute economy, and the knowledge from that error was gained after the first try; every try thereafter just resulted in people re-learning the same lessons as the first.

      • You use a computer to visit this website. In the 70s, we didn't even have computers that would be powerful enough to even display this website.

        No, I'm not using "a computer to browse a website", I'm using whatever cheap means for trivial entertainment is available to me, which just happens to be a cheap computer to browse a raunchy has-beemhas-been website. (I'm talking the general "me" here.)

        Back then they had a magnetophone and tapes, or theynhad a dirt bike and cheap gas, or a flea-infested cabin by the lake and spare time.on their hands to go fishing. All of those are less accessible, more expensive today than they were back then. That's why n

    • It's not a lie. You just have to include the overall effect on everyone's wages. CEO wage growth has been phenomenal.

    • Demand and wages is what's related. If you can provide a service, and nobody else is either willing or able* to provide that service at the price you're doing it for.

      *[they are unable to, due to anything from government/legal restrictions, didn't hear about the opportunity, are too far away, to lack of skill]

      A lot of our wages are simply because we're in the right place at the right time and the market is restricted. I mean, if I need a plumber to fix my sink it would be great if I could advertise that glob

    • by ranton ( 36917 )

      Productivity and wages are still related, but only productivity gains resulting from employees becoming more valuable. If productivity gains are coming from capital investment, like robotics and software, then the financial gains from productivity will go to capital (not labor).

      In the 50's and 60's, the populace in developed nations were becoming better educated. They became more valuable, and most of the productivity gains came from that. By the 70's and 80's, everyone who gains value from college arguably

  • by SuperKendall ( 25149 ) on Saturday February 18, 2023 @06:39PM (#63304505)

    What we have now has the trappings of being impressive because it's unexpected for people who have not been paying attention.

    I think when the actual abilities of AI may take off is more when AI can write better AI. Then you could potentially see massive leaps in ability. But, maybe even being that powerful it would just go off the rails and not produce more truly advanced AI in a way that is useful to us. Maybe a human will always have to be guiding.

    I would have zero worries about this as a new coder just entering the workforce, the main thing is to pay intention to it and from time to time try out AI assisted code generation tools - just as you should at all times pay attention to advancements in your field no matter what you are working on at the time.

    • I have already found a use for ChatGPT. I'm an example-based learner, and when I begin dealing with a new library or API, it can often (when it's not off the rails) produce sample code good enough to generalize from. I enjoy saving the "But do they mean it that way or this way? What experiment do I need to do to answer?" time drain.

    • by Moridineas ( 213502 ) on Saturday February 18, 2023 @07:33PM (#63304623) Journal

      I'm not worried, but I am using chatGPT daily to help with some rote tasks.

      Example: "convert this from Laravel 4.2 to Laravel 10" (paste in an entire file..done)

      Example: "write a javascript event handler for the click event of button btnLoad that uses a fetch to get the value of http://localhost/myscript [localhost], decodes the output as json, and emits a table with class "products" and one column from the outputted json name field, and one column from email field. Include full error handling with a console logging of errors"

      Works perfectly.

      • I'm not worried, but I am using chatGPT daily to help with some rote tasks.

        Example: "convert this from Laravel 4.2 to Laravel 10" (paste in an entire file..done)

        We're really going to be in trouble when we can say "ChatGPT, please look at these sections of a photo and indicate which ones contain a fire hydrant."

        • You jest, but I find the Laravel, and particularly Eloquent documentation to be absolute shit. Much easier to ask chatGPT.

    • When can AI improve itself? Since when has that been a concern when calling foreign level-1 support that barely speaks the native language? It's only been over a decade since IoT standardized the concept of absolute shit security Didn't seem to stop it from being sold and abused around the planet.

      Ain't it shocking how "improve" takes a back seat to profit. Every time. If we didn't give a shit about improving the humans we're replacing with AI, then maybe we should stop assuming AI has to be improved.

      TL

  • Coders will be first (Score:5, Interesting)

    by Kernel Kurtz ( 182424 ) on Saturday February 18, 2023 @06:41PM (#63304509)

    Programming is structured, algorithmic, and logic based. It should be much easier for a machine to do that than to have a deep philosophical discussion of anything.

    And of course once it can code well enough to code itself, then it can start to make improvements...

    • Necessity is the mother of all invention⦠I welcome tools to improve software, AI included. I don't expect strong AI to pop up soon. I recall watching a lecture from the 90s where one of the guys from Bletchley Park claimed strong AI would be commercially viable that decade. We joke about the two or three things being hard in computer science, but weâ(TM)d probably need a computer to help count how many ways it can be difficult, even with mastery.
    • by jythie ( 914043 ) on Saturday February 18, 2023 @07:11PM (#63304579)
      I picture it as a full circle thing. As AI gets better at programming, there will be a demand for people who can write specs to explain what they went the AI to do. Over time they will find that having a structured, formal language confuses the AI less, and COBOL rises from the dead for a bright new future, till spec writers decide COBOL sucks and rediscover C..
    • I see it the other way. Code is generally either right or wrong, there isn't much margin for error. Artistic/creative works have much more leeway. As such, given how today's AI systems work, i think they are likely going to be better at doing work that has that margin for error or multiple correct answers. It may be a useful tool for coders, but producing correct code without skilled supervision seems harder than generating clip art, stock photos, or creative writing.

      • I see it the other way. Code is generally either right or wrong, there isn't much margin for error. Artistic/creative works have much more leeway. As such, given how today's AI systems work, i think they are likely going to be better at doing work that has that margin for error or multiple correct answers. It may be a useful tool for coders, but producing correct code without skilled supervision seems harder than generating clip art, stock photos, or creative writing.

        Correct code has a reasonably well defined bar for correctness. There is much debate on this site about enforced memory safety in Rust vs C. An AI coder should always be memory safe in any language. This is iterative and trainable. It won't need the compiler to help it, it will be smart enough it does not need that.

        • Sure, it won't make errors that are easy to statically check for.

          But will it create race conditions?
          Data corruption?
          Code that falls over at scale?
          Subtly misinterpretted the requirements and ends up doing the wrong thing in edge cases?

          There are so many other ways for code to be fundamentally flawed that are not easily checked for or universally handled correctly in any existing code it could have "learned" from.

          • Sure, it won't make errors that are easy to statically check for.

            But will it create race conditions?
            Data corruption?
            Code that falls over at scale?
            Subtly misinterpretted the requirements and ends up doing the wrong thing in edge cases?

            There are so many other ways for code to be fundamentally flawed that are not easily checked for or universally handled correctly in any existing code it could have "learned" from.

            There are so many other ways for code to be fundamentally flawed that are not easily checked for or universally handled correctly in any existing code it could have "learned" from.

            Not saying it will be perfect, but it will learn way faster than humans. It won't be long before it is really only learning from its own code what is functional and what is not. The volume of outside example coding available just won't compete with the speed it can iterate and test variations of its own. It will teach itself much faster.

            • by djinn6 ( 1868030 ) on Saturday February 18, 2023 @10:46PM (#63304913)

              The current generation of "AI" won't be able to do that. They learn patterns and regurgitate those patterns. They don't have an internal logical analysis engine to check whether those patterns they're spitting out is actually correct.

              The search space for correct code is much larger than with something like chess or go. There's countless ways to write a short piece of compilable code that seems close to what you want, but to get code that is in fact correct requires the application of logic, not just patterns. It can keep trying different things till the stars all die and it won't have covered the entire problem space.

              Of course if your problem has already been solved elsewhere, the "AI" can regurgitate a solution for you. Not so much if it's a new problem. That's why they do so well for classroom and Leetcode-style questions. But they are alternatives to Google or Github, not alternatives to programmers.

              • AI could certainly put the best chess and go players out of business today if that was what we wanted. In any case AI only needs to outperform the average coder, not the ostensible coding grandmasters such as they may be, to be a threat.

          • It's for humans to look at the issues you rightly highlight and ensure that it does what it really should, as the humans should be at the moment. Unfortunately it doesn't happen now, and there's a serious danger it won't happen in the future.

    • Programming is not as logical as advertised. The value of the output is 100% subjective.
      • Programming is not as logical as advertised. The value of the output is 100% subjective.

        Output is a design decision, but I agree GIGO.

        • That too, but I'm also pointing out that most code doesn't actually do or mean anything, it's just birds chirping at each other.

          Personally, I think the world has wasted a lot of time and money on that while relatively few resources were devoted to mechanical operations.
    • Nope.

      Coding is about understanding a problem and finding a data model on the right level of abstraction. Most people can't even do that, machines will have it even mkre difficult.

      I'm guessing it'll drive down the need for dilettanti script kiddies, and what there will be left is demand for real programmers. Those who have to shell themselves short today because "the other guys do it for half the price", notwithstanding thatbthe other guys have no idea what they're doing. This'll become more obvious when the

    • ... structured, algorithmic, and logic based.

      Correction: 90% of reality is "structured, algorithmic, and logic based." Dealing with the remaining 10% requires experience of reality, of its edge-cases: Programming [source code] doesn't have that. I remind everyone, the first robot had to be programmed that gravity is real.

      Most of reality is difficult to explain: Why does Johnny want 1 car but 2 biscuits?

    • LLM are not logic based so I doubt it (another tech maybe but it doesn't exist currently). I recommend this link https://writings.stephenwolfra... [stephenwolfram.com] , chatgpt&co raise questions about what is human language, it's a lot more interesting than what they produce in fact.
    • Newsreaders, i.e. reporters, will be the first. It's pretty clear that the interviewer here was just barely capable of human speech.
    • Comment removed based on user account deletion
      • The mistake is thinking it is "just a chatbot". Give it another decade of learning and it will do much more than chat. Special purpose AIs other than ChatGPT are already being used to design drugs and solve physical problems that are every bit as complex as programming. Only a matter of time before general purpose AIs will reach that level of capability. I have great respect for developers, but I'm not sure it is harder than organic chemistry or nuclear physics, realms where some AIs can already do at le

  • Baffling (Score:5, Insightful)

    by locater16 ( 2326718 ) on Saturday February 18, 2023 @06:51PM (#63304537)
    Automation: Consistently eliminates jobs all throughout history.
    Journalists: "Will automation eliminate jobs in the future? This baffling and unanswered question will necessitate interviewing many self described experts."
    • No one doubts âoewillâ. The topic of the interviews and articles is âoewhenâ. And, yes, there are experts who can usefully weigh in to set probability ranges on that.

    • Re:Baffling (Score:4, Insightful)

      by upuv ( 1201447 ) on Saturday February 18, 2023 @10:20PM (#63304887) Journal

      Automation: Consistently eliminates jobs all throughout history.

      Well there are two ways to look at this statement.
      1. It's true
      2. It's false

      So how is it true. Automation will eliminate jobs for those with the skills that the automation directly addresses. Example: Monks in the middle ages copying books by hand. The printing press wiped out this type of job over night.

      So how is it false. Automation does two things. It lowers the price of the output of the automation. It creates new opportunities and thus jobs typically in vastly higher volume than those that were lost. The printing press opened up higher learning to the greater populous. It was the greatest uplift of almost all industries at the time. The work force in a short period of time made massive strides from almost entirely manual labour to one where more advanced industries could exist. The printing press enabled the industrial revolution more than any other technology.

      On a Net jobs equation. Every tech throughout the history of man has created a net jobs increase rather than remove them.

      Another example was the advent of "word processing" when the first PC's rolled off the assembly line the fear was that it would replace the office worker. Typing pools around the world feared for their job losses. And guess what. The typing pool did actually vanish. What replaced it. Well overall productivity of organisations went up. Transaction speed improved dramatically. A single person in a few minutes could now produce a document of detailed decision making, have it printed and sent off. A process that would take days when typing pools were used. This improved productivity led to a vastly expanded rate and volume of business.

      So what are the next opportunities that will be spawned from the assistance of chat GPT. Well if you figure it out let me know. Because I don't think anyone really understands which job functions are going to benefit most.

      Right now, people are afraid of change. All of the smoke coming out of people about what chatGPT is going to lead too is all just fear. This chatter about coders being impacted first is simply just fear based speculation. Could it be true? maybe. We are literally a few weeks into the public existence of the tool and people are losing their minds already.

      You want to be ahead of this curve. Find a niche where chatGPT will output results that are consistently accurate and of value. Capitalise on this as that you can sell this output faster than the next guy. Then find another two more niches. Rinse and repeat. Because this evolution will be rapid and very dynamic.

  • I'm at a university and work with faculty all the time. I really really REALLY want to see how an AI will work with faculty and not go completely off the rails. It's certainly something I have a hard enough time with.

    Or perhaps it'll be the faculty who go off the rails...

  • by Opportunist ( 166417 ) on Saturday February 18, 2023 @07:03PM (#63304553)

    Management. AI is predestined to replace managers. Managers' job is to make sure the correct amount of resources are allocated at the correct time at the correct place. That's trivial to do for an AI.

    Same for business decisions. These are based on economic models that require predictions of market developments. Another field that AI excels at.

    I really, really wonder why nobody ever had the idea to think in this direction. Could it be that it's only because nobody wants to eliminate themselves?

    • Management got replaced by case systems ages ago. You don't really need to pay somebody to make sure your employees are working you've got automated case systems that generate work for them to do and if the work isn't getting done or if an individual employee isn't doing enough of it you know.

      There are still people called the managers but their line workers who have been reclassified as managers either to trick them in the working harder for little or no pay increase or for lower tier jobs to enable wag
    • You don't need ChatGPT, a bash script can replace management.

    • Business decisions and managers are not the mass, it doesn't reduce cost to replace them. And obviously top managers won't replace themselves since its their business. Beside that, a lot of those jobs involve meeting people for business, it doesn't relate to chatgpt possibilities.
  • What about the middle managers?

    They seem to have the easiest jobs to replace.

    Passing along status and deciding how to balance workloads seem automatable even without AI.

    • I would think the layoffs indicate HR, Recruiting and Finance would be major ones. It could be the economic slow down, however I've seen a heavy use of AI in HR and Recruiting. While we've seen layoffs at various companies a surprising amount were not engineers that were laid off. Middle managers have been trying to justify return to office so they can be seen to do something. If it's remote we would see some predictable habits of managers monitoring their employees. HR needs to protect the company from emp
  • What!? (Score:5, Funny)

    by cygnusvis ( 6168614 ) on Saturday February 18, 2023 @07:09PM (#63304569)

    productivity and wages are related.

    HA!!!

    • they are not? I run a company and over time my company's productivity has gone up because of all of the capital I put into it. Also people working for me are getting increases once a year at least. Before I started it I was a contractor and hit the ceiling at some point so to go up from there I had to do more than just 10 hour work days, I had to increase my productivity and starting a company allowed me to do that. Now my productivity is much higher, before I could only sell my time to one client at a

      • You're the exception.

        What parent refers to is an official statistic that shows that the scissors between productivity and wages in the US is steadily increasing since the 70s. You can google, it's easy to find.

  • Lot of people did eye rolls on hearing his message just ONE election cycle ago.
    NB Just remember to vote for someone who won't kick you jobless ass to the gutter before you actually go homeless because you can't vote without an address.
  • The two issues are simply:

    1. Absolute individual share of total productivity. (In other words, how comfortable you are compared to ancestors)

    2. Relative individual share of total productivity, also known as individual political power, aka freedom. (Your ability to enact personal decisions for yourself when the most powerful individual in your society opposes them)

    The relationship between the two is not simple. Ancient Egypt was massive in #1, terrible in #2. A 19th century American farmer mig
  • As a professional software engineer?...definitely not. It is interesting. However, I'll worry once AI actually writes code. Show, don't tell. If Microsoft could write code with AI, it would. We wouldn't be asking if it could be done...we'd be evaluating AI-written applications right now. All the cash-rich big tech companies investing in AI have a lot of incentive to generate AI that could replace their arm of expensive engineers....or free them up so they can focus on new tasks.

    My take? It can't
  • Not likely. Remember, modern AI (machine learning) isn't really intelligence but a technical parlor trick supported by massive data sets. It does not think, it does not reason, and it is not aware. Long time AI researchers even see machine learning as a net negative to the field.

    So I have no fear of "AI" other than the massive amount of data being constantly mined from me and of corporations using ML as if it were actual AI.....I'm looking at you Tesla.

  • When Excel came out it eliminated a lot of jobs. It also created a lot of new jobs. AI is a tool, it still needs to be guided.

    If all you do is boilerplate coding, yes, your job will be eliminated. It was likely offshored already. There's always something new to learn though. If AI ever gets to a point where it can take all the computer jobs, then we'll either have UBI or mass riots.
  • by istartedi ( 132515 ) on Saturday February 18, 2023 @09:12PM (#63304807) Journal

    Maybe AI won't get pissed off at being called a "coder", but I always did. "Coder" is too low. "Software engineer" is too high. "Programmer" is just right, but nobody seems to use it any more. There's a computer. I program it. It's the most obvious choice, but NoooooO. "Learn to code"? Coding is what people do in the ER right before they die. Fuck that.

    You think an AI could write a paragraph like the above? Fat chance. I've seen the future, and it's pissed off. Humans are going to have a lock on that for a while, so get ahead of the curve and be angry now.

  • I noticed several places where EditorDavid apparently didn't notice who Gayle King is..
    For example:
          "...King tells his co-anchors..."
    and
        "...King synthesized what he'd learned."

    rather than "her co-anchors" or "she'd learned"

  • Hey ChatGPT, write me a code that will turn ChatGPT into a sentient being. ...

    Ladies and gentlemen, that is how Judgement Day shall begin.

  • by Beeftopia ( 1846720 ) on Saturday February 18, 2023 @11:08PM (#63304971)

    It's the same old thing: the consolidation of the production-of-value.

    That means value (i.e. things that people value) is created by fewer people using machines to increase their ability to create those things of value, be they goods (real or virtual) or services.

    When the Industrial Revolution happened, cottage industries like textile generation and subsistence farming were devastated. As production became centralized and generated more and more, in farming and manufacturing and with business processes, fewer people could generate more value.

    Here's a caution though: Pharmacists, lawyers, truck drivers have not been made obsolete by technology.

    Another caution is this, and it's a big one: Programming is hard and complex. Many people poo-poo programming. They think because they can do simple programming tasks, programming is easy. Any idiot can do it. Low barrier to entry. Programming is, however, like math and writing. Because you can add 2+2 doesn't mean you can do advanced mathematics. Because you can write "See dog run. Run dog run" doesn't mean you can write a novel or a college textbook. Look at how Musk rolled into Twitter and was a bull in a china shop, thinking his people would master the codebase effortlessly, but instead, breaking things.

    C has 32 keywords. The alphabet has 26 letters. There are 10 decimal digits and four key operators: addition, subtraction, multiplication and division. Just because you can do basic manipulation of these core character sets doesn't mean you can do advanced work with them. Linus Torvalds said that he was still working on the kernel because it's hard to program well. Linus is very, very smart and one of the top programmers in the world (Linux, git, etc).

    By the way, C++ has 95 keywords; Java has 51. Those are bigger character sets than the alphabet and there are a myriad ways to combine those keywords. And even then, think about what's happening: the words are translated into machine code and then translated into boolean algebra equations as they are run through the circuits of the processor. That's quite a bit of complexity.

    Is it possible that AI allows the generation of more value, like the tractor did for the farmer. But, because programming is like math or writing, I don't think AI will replace the human here.

    • by Arethan ( 223197 )

      me:

      tell me how chatgpt will be unable to replace software developers

      chatgpt:

      As an AI language model, ChatGPT can generate text, answer questions, and engage in conversations. However, there are certain limitations to what I can do and how I operate that make it unlikely that I will be able to replace software developers entirely. Here are a few reasons why:

      Lack of Creativity: While ChatGPT is capable of generating text, it is limited to the information it has been trained on. It can't come up with new and innovative solutions that require creative thinking. Software developers need to be creative to solve complex problems and come up with original solutions that fit a project's specific requirements.

      Limited Contextual Understanding: ChatGPT is great at processing and understanding language, but it has limitations in understanding the context in which software development takes place. Software development involves not just writing code, but also understanding the business needs, technical constraints, and user requirements. ChatGPT may not be able to fully comprehend these nuances and provide the same level of insight that a human software developer can.

      Inability to Code: ChatGPT is not capable of writing actual code. While it can generate text and suggest solutions, it cannot translate these solutions into actual software. Software development requires actual coding and debugging, which is something that ChatGPT is not currently capable of doing.

      Lack of Domain-Specific Knowledge: ChatGPT may have a vast knowledge base, but it may not have a deep understanding of specific programming languages or technologies. Software development requires expertise in specific programming languages, frameworks, and tools, which ChatGPT may not have the same level of knowledge and understanding.

      In conclusion, while ChatGPT is an impressive language model that can provide helpful insights, it cannot replace software developers entirely. Software development requires creativity, contextual understanding, coding expertise, and domain-specific knowledge, which ChatGPT may not be able to replicate in the same way as a human developer.

    • by qaz123 ( 2841887 )

      Another caution is this, and it's a big one: Programming is hard and complex.

      For humans. Moravec's paradox: https://en.wikipedia.org/wiki/... [wikipedia.org]

  • Ask ChatGPT. You will likely get a more accurate answer. If ChatGPT replaces pundits it will be a gain for humanity.
  • ...can we give them a login to Slashdot, so they can constantly post about how important it is to learn coding in every grade?

  • I mean we have gone from and assembling code to "problem oriented" programming languages, however it seems to me that actual productivity hasn't increased. It still takes a long time to write a simple database application. Something you could have clicked together in half an hour in Delphi now requires a programmer to actually sit down and write code.

    Having code generators powered by machine learning could, in theory, be yet another one of those potential productivity helpers.. yet none of the ones in the p

  • Haha. AI will eliminate jobs, alright. At places like CBS, which just refactors what others have said.

"Someone's been mean to you! Tell me who it is, so I can punch him tastefully." -- Ralph Bakshi's Mighty Mouse

Working...