Does Mike Rowe have some particular expertise in AI that would make his opinion more valuable than the thousands of other commentators saying it will ruin and/or save the world?
Doesn’t he make his money opining about how great blue collar jobs are and how “the kids these day” don’t want to do it. I’m not exactly surprised by his statement here.
That post is literally him shilling his merchandise.
>What to do then, in the face of such charlatans and double-dealers? What to do when Gavin Newsom tells me that masks must be worn for my own good, but refuses to wear one himself? That was question that led me to put “Safety Third” on cloth masks and start selling them to raise money for work ethic scholarships. I sold the masks with complete transparency. I told people I didn’t personally believe that a cloth mask would do anything to prevent the spread of an airborne virus, even if my governor did. I told everyone my Safety Third mask would keep you in compliance – but NOT out of danger. “This is a mask,” I wrote, “for anyone who wants to follow the rules, without being confused with those rule-followers who support mandates, or hypocritical politicians, or the elevation of safety above all else.”
So glad we have TV promoters to save us from the big bad doctors.
It's a particularly interesting contradiction for Mike Rowe, who has lamented the loss of skilled blue collar workers who can do the "dirty jobs", which robots certainly have not replaced yet.
Automation has made an impact on manufacturing, and will continue to, and AI will have an impact on knowledge work, but thus far technological leaps have lead to increased productivity rather than increased unemployment, although the changes in industries may cause displacements in the labor force. It seems logical to expect that this may play out similarly.
Dunno... I'm pretty AI positive, but after using ChatGPT 4 last night and having it hallucinate badly on a basic technical question I'm feeling much more confident I won't be replaced this year.
Whether or not ChatGPT can replace you is less immediately important than whether decision makers /believe/ it can replace you.
Organizations adopt non-productive practices for many different reasons and the consequences are often not felt for a long time.
Imagine a scenario where an executive decides that ChatGPT/CoPilot/Whatever makes a junior dev as productive as a senior. Even if they are wrong it could take years for the tech debt to build up enough for the organization to recognized it as a mistake.
>Even if they are wrong it could take years for the tech debt to build up enough for the organization to recognized it as a mistake.
This sounds like the weakest value proposition of your contribution ever. If you told me this was your greatest contribution over a junior with GPT I would take the junior with GPT any day. Things getting deployed for years before you start noticing the effects ? Most projects don't even last that long.
Anyway the current state of tech is that it can't really do that much outside of impressive demos and minor flow improvements, anyone taking a bet on that replacing people is going to lose big short term.
Just look at OpenAI - like I've said the other day, they have huge resources, can hire top talent, and all the infra they want at their disposal - they are still failing at solved problems (web app/payments), and the support they provide is worse than Google. Couldn't they dogfood a little having access to this tech far sooner than everyone else, with no restrictions and top talent available ?
And you're going to achieve that in an enterprise org that can't even hire mediocre talent and has no experience with AI ?
The only thing that's coming is crypto bro consultants and pundits, HODL AI too the moon !
My thesis: a combination of lower-cost deskilling and automation of labor management will have far more pernicious wage effects than higher-cost complete automation. Both white collar and blue collar skilled labor will be impacted.
A few other notes:
1. Lag Effects.
My father in law worked at an auto plant for most of his career and regularly pointed out that his job didn't exist at any of the newer factories. He continued working at that factory for over a decade after his position had already been automated on newer lines, and lost his job only when the factory was scrapped and replaced (and moved to a different state in the process, for tax benefits).
Robots destroyed a lot of skilled labor, em masse, over the last thirty years. And many of the jobs that it didn't already destroy will cease to exist over the next thirty years as new major capital investments are made and capital investments from 30+ years ago are wound down.
2. Automation/Optimization of Labor Management.
Over shorter time frames -- in-between large capital outlays -- automation of labor management ("Uberfication") is a lot more impactful than automation of labor. There are a lot of skilled jobs, both blue-collar and white-collar, which are now effectively commodified. Not automated, mind you, but definitely commodified. I think that's a much larger threat to plumbing and electrical work than automation.
3. Meta RE: Mike Rowe
Mike Rowe is a (very rich) media personality whose brand was built on calling attention to the trades as a viable path to the middle class and questioning the necessity of "four-year college for all". I think that's admirable.
Lately he's turned this into a wedge for separating two components of the non-managerial working class. I think that's deplorable. Software Engineers and Welders should be in solidarity, and as far as I can tell people like Mike Rowe function -- intentionally or unintentionally -- as lackeys for the ownership class.
IMO, the distinction between white collar and blue collar work is just red meat. Of course language models won't have any effect on plumbing or electrical work! But the idea that robotics won't de-skill a huge portion of blue collar work as well seems extraordinarily naive. (NOT automate. De-skill. Which, combined with automation of labor management, turns a career track into low-pay/no-benefit gigwork.)
My father was a plumber. A true trades/craftsman. Big new construction projects were deskilled with "installers" who were mostly under the table immigrants. The warranty claims on $800K became a cottage industry. Custom homes, commercial, and repairs still require skill and experience as far as I can tell.
> People used to say that the robots are going to destroy skilled labor.
Who was saying that? It was established (and has come to fruition) that robots would replace a lot of manual workers in unskilled jobs in places like factories. Much of this has been automated or outsourced. But we have millions of factory workers today that are working with robots and/or doing more complex manufacturing than in years past.
I also think there's a real 80/20 problem with LLMs right now. Having worked with them for a few months they are undoubtedly helpful, especially for simple things. And people will build useful and productive tools with them. But, in my experience, they only get you at best 80% of the way there - the remaining 20% just isn't there. Not yet anyways.
"The robots are coming for your job" is something that has been spouted for decades now. Pundits love to make these shortsighted predictions to get some air time.
I see AI as an enhancer, not a replacement, like computers or the Internet were before it.
I mean how many Horse Carriage Drivers are still around? Or blacksmiths?
Mike Rowe has a very good point, some fields will be losers and some fields will be winners. Some of those losing fields will be white collar workers so if you're in that losing field you should be prepared to career switch or become unemployed.
Obviously the headline is much more ominous that what he said.
> I mean how many Horse Carriage Drivers are still around?
Aren't there just drivers anymore?
> Or blacksmiths?
Metal workers are still a thing.
> Mike Rowe has a very good point, some fields will be losers and some fields will be winners.
He doesn't. For starters, he doesn't have a basic understanding of what AI does or means. In his mind, AI will replace workers because it is seen as autonomous, which definitely is not.
Though I find it ironic that AI could, theoretically, help replace TV show hosts like him pretty much tomorrow. We already have virtual hosts and AI could create both the character and the scripts.
I encourage you to try to drive a car the way you control a horse. It's a different set of skills; it'd be like saying that it's not problem AI will remove taxi drivers because then can just be a pilot. No, it will take retraining.
> Metal workers are still a thing.
Again, I'm not saying every job will disappear and neither is Rowe. There are still blacksmiths; there are just less of them.
> In his mind, AI will replace workers because it is seen as autonomous, which definitely is not.
I see you're intending to argue dishonestly and not actually look at his comments.
Happy that you agree with him. White Collar workers such as TV hosts may need to retrain into new jobs due to the advances in AI.
It seems to me that the problem is that the number of well paid fields seems to be decreasing and the difficulty of switching is prohibitive. E.g. I don't have the time or resources to go back to college in a new field and then rebuild a career from the ground up all over again.
This is the problem for me. I’ve been trying for years to find a way to go back to school, but I just don’t see how I can pay the ridiculous tuition + fees + rent + bills + every other necessity without working full time. And even then, I have a feeling I wouldn’t be able to complete a degree in the “usual” amount of time, which means if I did go back, I’d likely just end up old and competing with much fresher and probably much smarter people for a much lower salary. A friend straight up told me the other day the only reason they believed me and [another developer] could afford to live here is because we worked in software.
I was putting some savings together to try anyway, because the alternatives are status-quo or worse, but I’ve hemorrhaged a lot of those savings due to some emergencies + just being unemployed.
The usual approach is to attend school part time or online on nights and weekends while staying in your current job. If you only take one or two courses at a time this isn't a huge financial burden, especially if you can do some of it through a community college.
Yes, this is difficult and time consuming. It will require sacrificing other aspects of your life for several years. But many people find a way to make it work.
The way people work will certainly change. AI tools can or will soon be able to produce pretty good art, music, texts (both original and translations), etc. I think there will still be people working with these tools (to select, touch up, set the tasks, etc.), as well as more people producing such tools, but unquestionably the efficiencies created by these new technologies will result in a shrinking of headcounts in traditional areas.
As for the argument that professions like plumbers and electricians who physically visit their customers are not threatened by AI, this is not quite true.
For example, utility companies would send someone round to read our meter. Now they want us to have a smart meter (or, failing that, report our consumption online ourselves).
A current development in the automotive sector is toward systems that can be diagnosed and monitored automatically and remotely, and in some cases even fixed remotely, cutting out physical visits to a service centre. This will happen in other sectors as well.
Once your boiler or heat pump has enough sensors to monitor everything that could go wrong, physical visits will only be required once there is actually something that needs physically fixing. Etc.
As these things have become more technologically sophisticated over the decades, I've needed more frequent and longer visits with more failures than ever before. I'm not sure why that trend would be changing.
But we're talking about the labor market around the widget, not the TCO of the widget.
Automotive is a prime example: the added complexity increases lifetime maintenance costs, but at the same time the labor share of captured value has declined substantially.
I think this is currently the cardinal mistake people make when they that reason about the impact of automation. Labor doesn't get completely automated and replaced. Instead, one of two other things happen:
1. It gets de-skilled and commodified (GPS+Maps+smartphone -> complete deskilling of taxi driving).
2. Or it gets cut out from the profits (software in cars -> dealerships and "certified" mechanics capture more maintenance labor -> central entities get to extract more value. See also app stores.)
I just replaced the 30+ year old, non-intelligent water heater with a brand new non-intelligent water heater. And a 20 year old, non-intelligent air conditioner with the same. So if tomorrow morning, across the whole world, you can only buy AI-enabled, sensor ladened smart appliances, we will still need real people to cater to the poor dumb appliances for decades.
I don't understand your comment. Sam Altman has equity in OpenAI. It is in no way comparable to Mike Rowe's foundation which is a non-profit. He has no financial stake in it's success. He can't sell it for a profit.
Mike Rowe has accountants and attorneys to handle his tax filings. Complications are not a problem.
Not Engineering. Someone has to prompt the machine and check (almost always correct) its output. If you have copilot or codegen set up you know it's just better omnicomplete. This appears to be a fundamental limitation with language modeling.
Not Leadership, that requires humanity with all of the extra associated costs.
Maybe they could do some of the spam copywritting etc. Only really stuff that wasn't worth anything anyway. Businesses don't really seem to hire most White collar workers because their output is directly valuable though IMO. It's this super complicated social thing.
Sure for factory positions making mass produced welds... But that not what welders mostly do: robots have done those for a long time. A lot of welding is done in situ, out in the field, and a lot of it is one-off or few off welds that wouldn't make sense to automate even if you could get the robot into position. :)
And ignoring that, it's at least as useful as the brogrammer crowd bombarding journalists with "learn to code" a couple years ago was...
> And ignoring that, it's at least as useful as the brogrammer crowd bombarding journalists with "learn to code" a couple years ago was...
I think this is exactly the same as all these recent reports on AI.
Five years ago, that bootcamp trend made many people believe that they could earn six figures after taking six month courses. But that never happened, because there is more to becoming an effective engineer than just taking a few practical lessons. At the end, that was just marketing, but it didn't stop some managers from declaring that these high paying jobs would be soon gone, because of a massive influx of newly minted programmers.
The thing about engineers is that we have been automating boring stuff for ages. We will adapt to use AI tools, the same way we use auto-complete in IDEs, that is fine.
But AI will not replace engineers any time soon, and the same goes for the crowd who learnt to create websites in three month bootcamps.
