Hacker News Reader

Comment by runlaszlorun | Hacker News Reader

Everywhere??

That’s exactly the hype talk that’s going to burst this bubble.

Here’s tech’s dirty little secret. Despite all the screams about automation and universal basic income… the exact numbers where job replacement would show are in the labor productivity numbers. If GDP stays flat or grows while the number of jobs is reduced… bingo… you’d see that number climb.

Productivity has actually stayed flat or gone down over the last 15 years. Despite the fact that we’ve had trillion dollar corporate behemoths now. Despite that fact that we’re enabling a surveillance state Orwell couldn’t imagines. Despite the polarization we see. And teen anxiety going through the roof along teen/pre-teen suicides.

When you said AI (and in my view tech in general) are everywhere, I’m guessing this wasn’t what you meant…

View on HN

The missing productivity paradox is something that should interest everyone in tech! Like any macroeconomic observation, there is plenty of uncertainty in both the measurements (GDP ain't perfect) and the context (2 financial crises and a pandemic). But even pro-status-quo institutions like Brookings agree that Something Is Up. Their 2021 review on the topic is both a decent summary and a good source of further refs [1].

My favorite explanation is that many new technologies end up redistributing wealth rather than creating it, which certainly tracks with both subjective and quantified growth in inequality on the same time period. However, a slightly more optimistic take is that tech is aligning production better with people's preferences, so that the same productivity enables people to live more distinct lifestyles that suit them.

[1] https://www.brookings.edu/articles/how-to-solve-the-puzzle-o...

Another explanation that I find persuasive, put forward by Ezra Klein, is that productivity-sapping uses of technology grew along with the productivity-boosting ones: social media, for example, is a powerful mechanism for distracting people and destroying their attention spans.

If this is a good explanation, it begs the question of what AI might do to destroy productivity as well. If you’re constantly sexting with your AI girlfriend, who just happens to be extraordinary adept at tapping into your sexual proclivities, maybe you won’t get as many support tickets resolved as your boss was hoping.

More hypothetically, I would also expect that a world in which people spend a lot of time with screens strapped to their head, consuming an infinite stream of entertainment provided by generative AI, is not going to produce higher GDP.

> More hypothetically, I would also expect that a world in which people spend a lot of time with screens strapped to their head, consuming an infinite stream of entertainment provided by generative AI, is not going to produce higher GDP.

Yeah, I think this dovetails with the idea that IT may be satisfying preference allocations without increasing overall production. Watching 10 movies a month on a streaming service adds much less to the GDP than going to the cinema 10x, but if the selection is better it might satisfy you more. Economists sometimes attempt to measure this with "utility adjustments" which recognize increasing quality in the same goods, but it's very hard for those adjustments to account for the hidden preferences of the consumers as opposed to objective qualities of a good or service.

Information goods like social media and streaming, financial services like pay-me-later, and conveniences like next-day delivery are all examples of activities that might suit preferences without showing up in GDP. They also may enable distraction, waste, reclusiveness and impulsiveness in ways we'd like to avoid as a society. At the same time they might also help some people feel more included and less lonely or trapped by circumstance.

> put forward by Ezra Klein

The explanation sounds pretty familiar, so I might have already read/heard this from Klein, but would you mind sharing a link?

I believe he has mentioned it more than once, but one podcast that includes that discussion IIRC, and many other AI-related topics, is the April 7 episode “Why A.I. Might Not Take Your Job or Supercharge the Economy”. If you’re on iOS:

https://podcasts.apple.com/ca/podcast/the-ezra-klein-show/id...

It is less about productivity and more that AIs have the potential as the ideal employees.

No time off. No health care. Operate 24/7. No unions. No work safety concerns. No lawsuits over being unfairly fired. Control over exactly how something gets done or said.

If only the AIs will stop hallucinating or can consistently comply with policies …

> If only the AIs will stop hallucinating or can consistently comply with policies

I'm starting to wonder how much this matters.

People do crazy shit on the clock all the time. Company reps do not always adhere to policy 100% of the time either. People engage in office politics, coworkers accuse others of whatever, mistakes get made. LLMs happen to emulate all of this behavior.

In theory, we could replace everybody with AI and not much would be different. Productivity increase is debatable but cost savings would be immense. The question is how much insanity we're willing to tolerate as a result.

(...and seeing what fun ensues when there are more people than available jobs.)

>The question is how much insanity we're willing to tolerate as a result.

Given our elected representatives, I don't think that's a problem. No offense to any party or person. But we've consistently proven that we can tolerate and welcome way more insanity than seems reasonable.

How do those distinct lifestyles" match with rising inequality and people having a hard time buying a house or paying rent?

As far as I know, neither the rate of homelessness has gone up nor homeownership has gone down over the past decade.

https://endhomelessness.org/homelessness-in-america/homeless... It's increased.

The first one wouldn't cover rents increasing but people paying them, albeit struggling.

The second one is a lagging indicator. If lots of people bought their homes when they were cheap and are still alive, it will take time for the real impact to be visible.

Agreed. Though I think that the 'struggling' part creates a lag in both aspects. I think the 'gotta have a side-hustle' trend is a strong indicator of this. I would be interested in seeing the trends in number of people working 2+ jobs (or income streams), population shifts to more affordable areas, and number of disconnects of non-essential services. From experience growing up in a poor family, I know the definition of 'non-essential' can expand greatly as desperation grows.

Edit: Tracking real numbers in homelessness is also just extremely difficult.

> having a hard time buying a house

It's also much harder to get servants in that house. I wonder why...

I don't understand your comment.

if everything is so bad then it should be easier to find someone willing to work. But that's not the case today. Which means life today is probably not that bad.

Inequality is a red herring. Poverty is down.

Have you looked for housing in Columbus, Ohio?

I’m sure there are plenty of ghost towns with housing sitting empty that could be had for a song too. That probably doesn’t help if your job, family, and friends are somewhere else.

And remote work is more common than ever. Regardless, Columbus, Ohio has a very reasonable cost of living for the salaries available in the region.

Shhhh...don't need any more people here.

Ghost town?

https://www.intel.com/content/www/us/en/newsroom/news/intel-...

Nope. How am I supposed to get a job there? How about getting the money together to afford to move? I already live in a "low cost of living" area with less than half the population of Columbus, Ohio. It's still expensive as hell and there isn't a great housing situation.

Well, people are working everyday in Columbus OH.

I'm not from the US nor do I wish to live there, Columbus, Ohio or otherwise.

Is homelessness trending down or up? Why?

It's easier than ever to be homeless and it's generally frowned upon to forcefully institutionalize the mentally ill these days.

Wish I'd known that it was difficult to become homeless back when I was. It seemed really easy at the time, all you had to do was get evicted. Not sure how it could be easier now.

I think what they meant was that it's easier than ever to be homeless.

In many places you can't be thrown in jail for being homeless anymore. Many cities have more housing and shelters and free kitchens than ever before. Some places even give you a smartphone and basic plan. Etc.

My thought has been that most of the IT revolution hasn't been able to produce much extra goods, its all about Information, all it can do is help us optimize existing goods producing processes. As a result much of the productivity within the Information Technology advancement, has done little in the way of actual wealth creation, apart from optimizing existing processes, which hits a limit pretty fast.

> IT revolution hasn't been able to produce much extra goods

Measuring "Goods" in units or tons is bit simplistic. Almost everything is much better that it was 15-20 years back. TV, cars, phones, computers. This difference probably should be counted as 'extra', shouldn't it?

Your second paragraph makes intuitive sense to me. For every knowledge worker that got 2x as productive in the past decade thanks to new tech alone, there's a person who left their job where they were doing productive (perhaps grindy) things to do gig work because of the flexibility it provides. It might be nice burning VC money so someone can drive you home when you've had one too many (instead of taking public transit), and having someone shop for your groceries and walk your dog and do your laundry, but the individuals doing these tasks would probably accomplish more in terms of raw productive output if they were doing more traditional jobs.

Those are all traditional jobs though.

Driving, picking/packing, animal care, doing laundry - Absolutely nothing you mention is in any way some new 21st century job that didn't exist before. They're all just normal traditional jobs.

> When you said AI (and in my view tech in general) are everywhere, I’m guessing this wasn’t what you meant…

That person gave a list of tech they were talking about in their comment immediately afterwards: "speech to text, language translation, recommendation engines, relevancy ranking in search, computer vision, etc. and seems to be getting embedded in more and more processes by the day."

I'm not sure it's worth quibbling over whether we should use the term "everywhere" or "in many places"; the general point stands that it's found many different uses, and has done what effective tech does - fade into the background in many cases, just becoming part of our daily lives.

Sure, we're not seeing the off the wall predictions from the singularity crowd, but it seems to be tech that most people find broadly useful.

I've seen some indication that this is a measurement problem. Tech enables computers to do things that would be originally expensive for humans to do.

Translation is an interesting example: some labor has been displaced, but not nearly all because there's still value in having human eyes carefully checking the translation of high value documents. But free translation let's regular people translate things freely - a new capability which displaced no one.

However, productivity measures human productivity and human labor. The very cheap new translation modality is therefore completely missed by productivity measurements.

Meanwhile, there's /more/ jobs available right now, despite all of this. The US has hit a historic low in unemployment and wages are going up, leading to a decline in measured productivity. Productivity is output per dollar of wages, which means we twist our hands in anxiety when workers start doing better...

> GDP stays flat or grows while the number of jobs is reduced

I lack the economics knowledge to do more than parrot the response I've heard to this, so take this with the appropriate level of "hmmm":

As I understand it, the counter-claim is that the measure of GDP mostly excludes exactly the set of things that grows absurdly fast.

For example, the measure of inflation may include the cost of a smartphone in the standard basket of goods, but not the fact the GPU of a smartphone (or Apple TV) of today, operating in double precision mode, can do more than the Numerical Wind Tunnel supercomputer in 1993 costing 100 million dollars.

Or that everyone has a free encyclopaedia a hundred times the size of the Encyclopædia Britannica.

And maps which for most users are as good as Ordnance Survey, but free and worldwide, when the actual OS prices for just the UK is… currently discounted to £2,818.17, from £4,025.97.

Or that getting you genome sequenced now costs a grand rather than 3 billion. Although that might not yet even be in the basket, I don't know where the actual baskets of goods get listed in most cases, and search results aren't helping — one result, on a government website, lists "health", but even digging into the spreadsheet didn't illuminate much detail there.

That is true. Switching from buying Britannica to using Wikipedia is counted as a reduction in GDP as GDP counts what you spend and Wikipedia is free, even if it's better.

The UK basket of goods is here https://www.ons.gov.uk/economy/inflationandpriceindices/arti... and the various sublinks.

Good. GDP should not include any of those things. Those are tools. GDP includes outputs and impacts.

Maybe you design a wrench that is 1000x cheaper and faster to use and more reliable. Well, if it makes your car building operation 0.0001% faster, that's the impact. The details of the wrench and how impressive it is are irrelevant to any observer.

If having your genome sequenced leads to far longer or better lives then we would see the impact in productivity. Same with everything else on the list.

> …genome costs…

Source: This week’s Kurzgesagt, right?

Not only but also; this was a thing I was aware of this years back, but Kurzgesagt is a nice bit of easy watching when I'm eating dinner.

Why does GDP measure anyway? How much money do we create for billionaires?

My life is permeated by tech (and big part of it is AI) and made 100 times easier. I can buy a plane ticket to another country while waiting for a subway (did people use to take an hour to go to a special place, wait in line and talk to human to buy a ticket? I still remember this). I go there and quickly navigate in a city I know next to nothing about, find something niche, like cool local cafés in the area because GPS and google maps. I go to a restaurant and I can use google translate to understand the menu. I don't even need to type unfamiliar words, AI scans the image and translates it on the fly. The same google translate with speech recognition AI helps me to converse with a person when we don't share any common language. I can click couple of buttons and video-call my mum who lives on the other side of the world. If I need to buy something I need very rarely, I can order it online and not think where I find a shop that sells those things. Even if I don't know the right word, I can now ask chatgpt "what do you call in German that fancy thing you mount on the ceiling and attach lights to it?"

My life is _hugely_ more efficient thanks to tech and AI. Does it help me to contribute more to the abstract economic growth? I don't know, perhaps not. But I just don't care about GDP.

GDP is not productivity. If I manage to produce/sell a hundred gizmos for $1 each, displacing my competition who were producing/selling 50 gizmos for a price of $5 each, I just halvened gdp produced even though both me and people using the gizmo are more productive than before.

But your customers now have money to spend on something else. The money didn’t disappear.

I think precisely the point is that money isn't wealth.

Government can print as much of the latter as they want; wealth goes up when we can collectively buy more and down when we can collectively buy less, regardless of how many dollarpounds that is.

But GDP is measured in money, and can only connect to wealth if we get get inflation right, but that's really hard because inflation depends on what you want to buy — childcare costs don't matter if you have no kids of that age.

That said, I trust the domain experts to get this right, even though the various governments may be incentivised to claim their own preferred numbers. Even at worst, they'll have thought of vastly more influences than I can even imagine.

Maybe you live in a monolinguistic bubble, or speak every language fluently, but when is the last time you hired a human being to translate a language you don't understand instead of using AI like Google Translate?

AI is already so ubiquitous and useful that you blindly take it for granted without even thinking.

Most companies will at least hire a human to check translation work, or at least contract one. Bad translations are something that can quickly destroy a brand in a market, and translation tools are not so fine grained to take into account dialect.

Spanish, for example, has many examples of words that are innocuous in one dialect and profane in another.

> Bad translations are something that can quickly destroy a brand in a market

Bad translations are present in product names & descriptions of at least 70% of all products on Amazon and Ebay I've seen, and it doesn't look like it hurts the business in any way.

Have you seen their sales revenues?

Big difference between "buy this $5 plastic crap widget despite its product description being barely coherent because there's only so many seconds you're willing to spare searching for 'Plastic Crap Widget' because Familiar Tech Company's algorithm puts CBSPOO ahead of VRIENGLU in this particuar search parameter combo" and "hello professionals in niche market. You will recognise the name of our product whenever you next have $xx,xxx to spend on these services, because we were the one whose ad was inadvertently extremely sexist"

I think that people who are fine with buying crappy products don't care about crappy translations.

Others, however, do care. I, for one, will pass on something that has bad translation because I take that as a proxy for the quality of the product overall.

> when is the last time you hired a human being to translate a language you don't understand instead of using AI like Google Translate?

Last year, for an important letter that had to be written in Japanese, a language that I don't know. Using Google Translate for that was unthinkable, because Google Translate is pretty poor and I had no way of checking and correcting the translated text.

Dealing with any international company requires having a translation office with a distinguishable stamp on every single letter. Might be slow, but they can be held accountable unlike your number soup AI

Especially in translation quality. Being able to run high quality translation model that does 100 languages directly without going through English on a desktop pc with a previous gen gpu (2070 in my case) is huuuuge. (I'm talking about fairseq and m2m100 for anyone interested in spinning up their own).

If I was going to name the biggest good thing AI has done to humanity so far is the ability to read internet sites in other languages like Chinese (Google sucks at it, you have to use other tools, I use an app called "tap translate screen"). Also ability to do voice to text and translation at the same time on mobile devices (currently requires online connection).

AI affects both human interfaces (text, speech, sound, images) and the ability to interpret and create data by/for humans. That's effectively the entire interface space between machine and the human-experienced world. Can you think of an area of technology that won't be affected by AI? Even unrelated technologies are going to get affected in their interfaces and tooling.

As for the rest of your comment... please don't hijack other conversations for soapboxing on the industry as a whole. Instead, submit your post and open a real conversation.

Reply on news.ycombinator.com