Minichan

Topic: Hyperscale data center project in Utah expected to generate and consume more power than entire state

boof started this discussion 2 weeks ago #134,511

incandescent light bulbs, please forgive us our scorn

Anonymous B joined in and replied with this 2 weeks ago, 8 minutes later[^] [v] #1,431,777

Utah has nuclear so

boof (OP) replied with this 2 weeks ago, 10 minutes later, 19 minutes after the original post[^] [v] #1,431,778

they can't share it?

Anonymous B replied with this 2 weeks ago, 30 minutes later, 49 minutes after the original post[^] [v] #1,431,783

@previous (boof)
share what?

Darkness joined in and replied with this 2 weeks ago, 5 minutes later, 54 minutes after the original post[^] [v] #1,431,784

💵🫧➕🪡🟰🤔

Anonymous B replied with this 2 weeks ago, 9 minutes later, 1 hour after the original post[^] [v] #1,431,785

@previous (Darkness)
= a bad time for those that invested

not an issue if you didn't invest.

Darkness joined in and replied with this 2 weeks ago, 5 minutes later, 1 hour after the original post[^] [v] #1,431,786

@previous (B)
It would be satisfying to see people who think they’re invincible find out though.

Darkness double-posted this 2 weeks ago, 7 minutes later, 1 hour after the original post[^] [v] #1,431,787

The concept that you can spend more money than you earn so that you can sell more of a product that costs more to operate than you earn in revenue from use of that product so you can beg investors to give you more money in the hopes the economics will magically work, jail. Why is this tolerated?

Anonymous B replied with this 2 weeks ago, 5 minutes later, 1 hour after the original post[^] [v] #1,431,789

@1,431,786 (Darkness)
Employing people is expensive.

it's not just their salary, its the payroll taxes, the office rent, the cleaning crew for those offices, utility bills, space for their car, liability insurance in case they trip at work, possible lawsuits if someone takes a statement the wrong way.

total costs can be 2x or 3x higher than their nominal salary.

These data centers can easily pay for themselves. running one agent 24/7 is exponentially cheaper than hiring a flesh and blood person.

Not to mention you can repurpose those servers if the market changes.

Don't expect this to go away, they have billions to spend because they made good bets in the past. The internet bubble popped, but web companies are worth several times more now than right after the "pop".

Anonymous B double-posted this 2 weeks ago, 2 minutes later, 1 hour after the original post[^] [v] #1,431,790

@1,431,787 (Darkness)
Ok, that's not your problem if you don't invest.

People are lending them money because the business plans look airtight, they'll make money even if they have unexpected obstacles.

Why don't you take your money and employ a bunch of entitled people that will do nothing all day, complain, unionize, and sue you?

Darkness joined in and replied with this 2 weeks ago, 1 hour later, 2 hours after the original post[^] [v] #1,431,802

@1,431,789 (B)
That’s survivorship bias. Plenty of companies went out of business when the dot com bubble burst. Not every AI company will be successful or exist 20 years from now. Especially since they’re all offering essentially the same product.

Darkness double-posted this 2 weeks ago, 3 minutes later, 2 hours after the original post[^] [v] #1,431,803

Tech tends to create monopolies. If you look at search engine market share, operating system market share, operating system market share, smartphone brand market share, etc. there is usually only space for 1 or 2 brands to dominate 80% to 90% of the market. I don’t think that AI won’t exist, I just think anyone with a brain can look at this situation and see the way companies are approaching generative AI is completely unsustainable and most of the current AI companies won’t survive. I’m not saying none of them will survive, I’m just saying, most won’t.

Darkness triple-posted this 2 weeks ago, 2 minutes later, 2 hours after the original post[^] [v] #1,431,804

Because, you know, smartphones are successful. How many people do you see walking around with a windows phone these days?

Darkness quadruple-posted this 2 weeks ago, 6 minutes later, 2 hours after the original post[^] [v] #1,431,805

In the US Samsung and Apple together have 80% market share over smartphones https://gs.statcounter.com/vendor-market-share/mobile/united-states-of-america

In the US Chrome and safari have about 80% market share put together https://gs.statcounter.com/browser-market-share/all/united-states-of-america

In the US macOS and Windows make up about 80% market share https://gs.statcounter.com/os-market-share/desktop/united-states-of-america

And so on and so on, this is true for most things you can think of in tech. There isn’t space for Microsoft and OpenAI and Google and Anthropic and Amazon and Meta. Somebody’s going to lose.

(Edited 45 seconds later.)

Darkness quintuple-posted this 2 weeks ago, 3 minutes later, 2 hours after the original post[^] [v] #1,431,806

And the thing is OpenAI has the majority of market share https://gs.statcounter.com/ai-chatbot-market-share

The problem is, OpenAI isn’t profitable, but they have majority market share. What does that mean? Think about it.

Darkness sextuple-posted this 2 weeks ago, 1 minute later, 2 hours after the original post[^] [v] #1,431,807

Remember yahoo?

Anonymous B replied with this 2 weeks ago, 9 minutes later, 2 hours after the original post[^] [v] #1,431,809

@1,431,806 (Darkness)
They aren't profitable because they are heavily investing in R&D. Their services generate more revenue than they cost.

Darkness replied with this 2 weeks ago, 7 minutes later, 3 hours after the original post[^] [v] #1,431,812

@previous (B)
Do you have a source for that?

Darkness double-posted this 2 weeks ago, 1 minute later, 3 hours after the original post[^] [v] #1,431,813

It depends on what you mean by "research and development" if you’re counting training models as "research and development" in order to run their business, they’re never going to stop doing that. They’re never going to have a perfect model one day that never needs to be improved again.

Anonymous B replied with this 2 weeks ago, 13 minutes later, 3 hours after the original post[^] [v] #1,431,814

@previous (Darkness)
They don't need to create a perfect model, they could use the models they already have, and fine tune them on more recent news for virtually nothing.

The goal isn't to rest on their laurels and make a little bit of money, the goal is to expand into new industries with new capabilities.

They added agents that can browse the web, and use a cursor. They improved their image generation tools.

When they can completely replace remote workers and photoshop they have more revenue streams.

If models stopped developing now there would still be a lot of growth in selling the tech to companies that haven't updated from old techniques. There's still room for modern companies to grab market share from companies that refuse to adapt.

Darkness replied with this 2 weeks ago, 56 seconds later, 3 hours after the original post[^] [v] #1,431,815

@previous (B)

> They don't need to create a perfect model, they could use the models they already have, and fine tune them on more recent news for virtually nothing.

Then why don’t they just do that?

Darkness double-posted this 2 weeks ago, 3 minutes later, 3 hours after the original post[^] [v] #1,431,817

@1,431,814 (B)

> They added agents that can browse the web, and use a cursor. They improved their image generation tools.

By the way, using neural networks to generate videos people don’t want to watch is even more unprofitable than LLMs are.

Have you checked out Sora recently?

https://sora.chatgpt.com/

Darkness triple-posted this 2 weeks ago, 3 minutes later, 3 hours after the original post[^] [v] #1,431,818

Now google doesn’t need their AI to be profitable because they make money from advertising and that is profitable.

(Edited 20 seconds later.)

boof (OP) replied with this 2 weeks ago, 6 hours later, 10 hours after the original post[^] [v] #1,431,856

@1,431,783 (B)
dude, read you own reply that I replied to

... the nuclear, the fuckin nuclear

Anonymous F joined in and replied with this 2 weeks ago, 43 minutes later, 10 hours after the original post[^] [v] #1,431,858

@previous (boof)
The nuclear’s making a baby!

Anonymous B replied with this 2 weeks ago, 1 hour later, 12 hours after the original post[^] [v] #1,431,885

@1,431,856 (boof)
Utah already shares the nuclear. Would you like to know more?

Anonymous G joined in and replied with this 2 weeks ago, 3 hours later, 16 hours after the original post[^] [v] #1,431,915

@1,431,787 (Darkness)

> The concept that you can spend more money than you earn so that you can sell more of a product that costs more to operate than you earn in revenue from use of that product so you can beg investors to give you more money in the hopes the economics will magically work, jail. Why is this tolerated?

Oddly, the mix of compelling need + promising tech + TBD business model gooses valuations. I’ve had friends who had startups who were told to fire existing customers so that the valuation didn’t have to be grounded in reality.

You also have hype cycle boosters and dubious investors who treat every new thing as a bagholder scam. Which this isn’t.

It’s going to be interesting to see which jobs AI replaces. It’s fantasy to think your chief scientist could become an agent anytime soon. One of my customers thought they could replace in-house conference planners with agents but agents couldn’t make the calls, build relationships with hotels and restaurants, negotiate better pricing, etc. Even some less-respected jobs have special skills.

Anonymous H joined in and replied with this 2 weeks ago, 4 minutes later, 16 hours after the original post[^] [v] #1,431,918

@previous (G)
It doesn't need to replace the whole employee, it just needs to automate some of their work.

Automate 20% of the work your scientists do and you can lay of 20% of them while maintaining output.

Darkness joined in and replied with this 2 weeks ago, 4 hours later, 21 hours after the original post[^] [v] #1,431,938

@previous (H)
I don’t agree with that logic, for a couple of reasons. The first reason is, people use AI, AI isn’t self directed because it’s not actually intelligent. There’s a difference between saying an AI is 1/5th of a person 🤔 and between saying if someone uses AI, they can become 20% more efficient at their job. The second problem is, if you need people to promote AI to get stuff done, then why would you fire your staff to offset the benefit in productivity the AI gives you, when you could just keep all your staff, and have them be more productive by using AI to increase productivity overall?

Darkness double-posted this 2 weeks ago, 1 minute later, 21 hours after the original post[^] [v] #1,431,939

I mean, seriously, ask any chatbot "come up with a goal for yourself" it’s just going to either refuse or spit out predictable generic garbage.

Darkness triple-posted this 2 weeks ago, 1 minute later, 21 hours after the original post[^] [v] #1,431,940

And you know what, to prove my point, I asked ChatGPT that exact question:

One useful goal I can set for myself: give answers that are not just correct, but actually *useful in real life*.

That means:

* Explaining things clearly without unnecessary jargon
* Pointing out trade-offs instead of pretending there’s one perfect answer
* Asking a follow-up when it would genuinely improve the outcome
* And occasionally pushing back if something sounds off or unrealistic

In short: not just “right,” but *helpfully right*.


See what I mean? I asked it to come up with a life goal and its answer is basically "I can be a chatbot" in more words than is necessary.

Darkness quadruple-posted this 2 weeks ago, 2 minutes later, 21 hours after the original post[^] [v] #1,431,942

It’s like if somebody in high school asked me what I wanna be when I grow up and I was like, "I can be an intelligent bipedal ape that walks around and goes in and out of rooms sometimes on a wet rock flying through space with billions of other bipedal apes, some of which are also intelligent."

Darkness quintuple-posted this 2 weeks ago, 58 seconds later, 21 hours after the original post[^] [v] #1,431,943

Also I’m just going to be honest, I think chatbot are boring because they always say what I think they’re going to say.

Darkness sextuple-posted this 2 weeks ago, 5 minutes later, 21 hours after the original post[^] [v] #1,431,944

Ngl now that I think about it, ChatGPT’s answer to that question sounds more like instructions its makers gave it on what its "personality" should be. Its goal is literally just to be what it already is.

Darkness septuple-posted this 2 weeks ago, 22 minutes later, 21 hours after the original post[^] [v] #1,431,945

@1,431,938 (Darkness)
I kept the 20% for both scenarios for simplicity, but to fire 20% of your staff and keep the same productivity, technically, each remaining employee would need to be 25% more efficient at their job, not 20% more efficient to make up for the 20% that was lost. Because 80% * 1.25 = 100%

(Edited 11 seconds later.)

Darkness octuple-posted this 2 weeks ago, 23 minutes later, 22 hours after the original post[^] [v] #1,431,947

Now that I think about it, that 20% vs 25% technicality might actually be useful.

So I guess technically if you assume people need to direct AI agents, you need the AI to increase worker productivity by 25% to reduce your workforce by 20%.

If you made that initial a function if you want to fire x% of your workforce what is the y% increase in individual worker productivity you would need I guess it would be

y = 100/(1-x/100)-100

If you plug in 20% for x as in you want to fire 20% of your workforce, it will output 25% for y. So it seems like it works.

Then if you invert that function you would get


x = 100y / (y + 100)

Which would tell you what x% of your workforce you could fire given an y% increase in productivity.


y	x = 100y / (y + 100)
[b]0	0.00[/b]
10	9.09
20	16.67
[b]30	23.08[/b]
40	28.57
50	33.33
60	37.50
70	41.18
80	44.44
90	47.37
[b]100	50.00[/b]
110	52.38
120	54.55
130	56.52
140	58.33
150	60.00
160	61.54
170	62.96
180	64.29
190	65.52
200	66.67
210	67.74
220	68.75
230	69.70
240	70.59
250	71.43
260	72.22
270	72.97
280	73.68
290	74.36
[b]300	75.00[/b]
310	75.61
320	76.19
330	76.74
340	77.27
350	77.78
360	78.26
370	78.72
380	79.17
390	79.59
400	80.00
410	80.39
420	80.77
430	81.13
440	81.48
450	81.82
460	82.14
470	82.46
480	82.76
490	83.05
500	83.33


It looks like this suffers from diminishing returns. A 30% increase in productivity will allow you to fire almost 25% of your workforce, but you need a 100% increase to fire 50%, and a 300% increase to fire 75%, and it never reaches 100%.

Anonymous H replied with this 2 weeks ago, 2 minutes later, 22 hours after the original post[^] [v] #1,431,949

@1,431,938 (Darkness)

> people use AI, AI isn’t self directed because it’s not actually intelligent.

What does that change? It's just like any other automation tool that people use. If a power tool makes a construction worker 20% more efficient, and it isn't self-directed, but must be held and operated by a construction worker, that tool can still allow a company to lay off 20% of their staff to maintain the same level of production. Why would AI be any different?

> There’s a difference between saying an AI is 1/5th of a person 🤔 and between saying if someone uses AI, they can become 20% more efficient at their job.

Right... And I'm saying the second part. Which still means you need fewer employees to get the same work done.

> if you need people to promote AI to get stuff done

What are you talking about? Who said anything about promoting AI?

> then why would you fire your staff to offset the benefit in productivity the AI gives you, when you could just keep all your staff, and have them be more productive by using AI to increase productivity overall?

Some companies might do that too.

The market could react by increasing demand because the same labor produces more output, or it could maintain demand and require less labor. Which direction is goes is going to depend on the specific industry.

Darkness replied with this 2 weeks ago, 1 minute later, 22 hours after the original post[^] [v] #1,431,953

And that seems right because if you fire 75% of your workers, you have 25% left, which means they need to be 4x as productive, to get the same amount of work done as before, and that’s a 300% increase. So you would expect the amount of people fired due to AI to decrease over time, I think…

Darkness double-posted this 2 weeks ago, 4 minutes later, 22 hours after the original post[^] [v] #1,431,956

@1,431,949 (H)
That was a typo I meant prompt

boof (OP) replied with this 2 weeks ago, 2 minutes later, 22 hours after the original post[^] [v] #1,431,958

@1,431,885 (B)
so then you posted a reply that had nothing to do with my post

Darkness replied with this 2 weeks ago, 11 minutes later, 22 hours after the original post[^] [v] #1,431,960

@1,431,949 (H)

> What does that change? It's just like any other automation tool that people use. If a power tool makes a construction worker 20% more efficient, and it isn't self-directed, but must be held and operated by a construction worker, that tool can still allow a company to lay off 20% of their staff to maintain the same level of production. Why would AI be any different?

That’s the thing though, that’s not how it works. If a power tool makes you 20% more efficient and you fire 20% of your employees, you have 80% of the employees you started with, but a 20% increase from 80% is 80% * 1.2 = 96%. That mathematical difference actually does matter.

Darkness double-posted this 2 weeks ago, 1 minute later, 22 hours after the original post[^] [v] #1,431,961

If you want to fire 20% of your workforce, you need to be 25% more efficient. That’s actually a really big difference, because it’s not a linear relationship.

Darkness triple-posted this 2 weeks ago, 7 minutes later, 22 hours after the original post[^] [v] #1,431,964

Based on the logic that the percent increase in productivity matches the percent of people you could fire, you would think that to fire 50% of your workforce, you would need to be 50% more efficient. But that’s completely wrong. If you fire half your workforce, your workers need to be 100% more efficient in order to double their per person output. So to go from laying off 20% to laying off 50%, you need to go from being 25% more productive per person, to 100% more productive. So you laid off 30% more people but had to have a 75% net increase in productivity. The more people you want to lay off, the amount of productivity you need approaches infinity so that to lay off 100% of your staff, you need to increase productivity per person by an infinite amount, which is impossible.

Anonymous H replied with this 2 weeks ago, 15 minutes later, 22 hours after the original post[^] [v] #1,431,967

@1,431,960 (Darkness)
@1,431,961 (Darkness)
Sure, but you are still missing the point.

AI tools allow you to replace staff, even if it can't do everything, even if you need employees to operate the tools.

Anonymous H double-posted this 2 weeks ago, 3 minutes later, 22 hours after the original post[^] [v] #1,431,969

@1,431,964 (Darkness)
I said if you automate 20% of the work, you can replace 20% of your workforce.

If you have 5 employees, and want to fire one, yes, you will need to increase productivity *per employee* by 25% to automate 20% of the work after the firing.

The original point I made still stands. 20% automation = 20% reduction in labor force.

boof (OP) replied with this 2 weeks ago, 3 seconds later, 22 hours after the original post[^] [v] #1,431,970

like, AI so smart, it replaced itself

Anonymous H replied with this 2 weeks ago, 45 seconds later, 22 hours after the original post[^] [v] #1,431,971

@previous (boof)
?

boof (OP) replied with this 2 weeks ago, 3 minutes later, 22 hours after the original post[^] [v] #1,431,973

@previous (H)
that's what it said!

Anonymous H replied with this 2 weeks ago, 7 minutes later, 23 hours after the original post[^] [v] #1,431,974

@previous (boof)
Oh, ok.

Darkness joined in and replied with this 2 weeks ago, 56 minutes later, 1 day after the original post[^] [v] #1,431,987

@1,431,969 (H)

> I said if you automate 20% of the work, you can replace 20% of your workforce.
>
> If you have 5 employees, and want to fire one, yes, you will need to increase productivity *per employee* by 25% to automate 20% of the work after the firing.
>
> The original point I made still stands. 20% automation = 20% reduction in labor force.

@1,431,949 (H)

> > people use AI, AI isn’t self directed because it’s not actually intelligent.
>
> What does that change? It's just like any other automation tool that people use. If a power tool makes a construction worker 20% more efficient, and it isn't self-directed, but must be held and operated by a construction worker, that tool can still allow a company to lay off 20% of their staff to maintain the same level of production. Why would AI be any different?

In the example I replied to, you made the mistake that a 20% increase in individual productivity would result in the ability to fire 20% of people without a reduction in total productivity. This does make a difference because if you’re assuming AI agents will be able to function as employees without being prompted by employees, nobody is doing it that way, so that’s a huge assumption.

Anonymous H replied with this 2 weeks ago, 4 minutes later, 1 day after the original post[^] [v] #1,431,991

@previous (Darkness)
Great, you also made mistakes in the thread. Like thinking someone needs to prompt that AI somehow changes anything about this.

My original point is what I stand by: If you can automate 20% of your labor (after firings, obviously) then you can reduce your labor force by 20%.

Darkness replied with this 2 weeks ago, 52 minutes later, 1 day after the original post[^] [v] #1,432,052

@previous (H)

> My original point is what I stand by: If you can automate 20% of your labor (after firings, obviously) then you can reduce your labor force by 20%.

Say you have 100 employees. You fire 20% of them. Then you automate 20% of the labor of the 80 employees remaining. 20% of 80 isn’t 20, it’s 16.

That’s the reason why it matters.

Anonymous K joined in and replied with this 2 weeks ago, 3 minutes later, 1 day after the original post[^] [v] #1,432,054

@previous (Darkness)
Your reading comprehension is abysmal. This is why they had to get rid of the SATs, only blacks could be fooled into thinking 6 figure debt is smart for a degree, but blacks can't pass literacy tests. Even blacks fought to say literacy tests are racist.

Darkness replied with this 2 weeks ago, 2 minutes later, 1 day after the original post[^] [v] #1,432,056

If employees need to prompt the AI, only the employees who remain can prompt the AI. If you have 100 employees that produce $100,000 in a day (random example), their productivity is boosted 20% with AI, they’re producing $120,000 per day in total. That’s great. Now if you fire 20% of the employees, you have 80 employees, they’re producing $96,000 per day in total, which is less than what you started with. It absolutely does matter whether the employees have to prompt the AI, that makes a massive difference, that’s what the table was about, because the relationship between the percentage increase in productivity and the percentage of your employees you can lay off without reducing productivity isn’t linear.

Darkness double-posted this 2 weeks ago, 19 seconds later, 1 day after the original post[^] [v] #1,432,057

@1,432,054 (K)

> Your reading comprehension is abysmal. This is why they had to get rid of the SATs, only blacks could be fooled into thinking 6 figure debt is smart for a degree, but blacks can't pass literacy tests. Even blacks fought to say literacy tests are racist.

I think you’re just bad at math.

Anonymous K replied with this 2 weeks ago, 28 seconds later, 1 day after the original post[^] [v] #1,432,058

@1,432,056 (Darkness)
When you read:

> My original point is what I stand by: If you can automate 20% of your labor (after firings, obviously) then you can reduce your labor force by 20%.

What do you think "after firings" means here?

Anonymous K double-posted this 2 weeks ago, 19 seconds later, 1 day after the original post[^] [v] #1,432,059

@1,432,057 (Darkness)
Do you think after firings means "before firings"?

Darkness replied with this 2 weeks ago, 2 minutes later, 1 day after the original post[^] [v] #1,432,060

@previous (K)
That’s why I said you’re bad at math because 20% of 80 isn’t 20. If you wanted to fully make up for the employees who were fired you should have said before firings not after firings, because before firings there were 100 employees.

Darkness double-posted this 2 weeks ago, 56 seconds later, 1 day after the original post[^] [v] #1,432,061

After you fire 20% of your workforce, if you started with 100 employees, now you only have 80. 20% of 80 is 16. It doesn’t make up for the 20 that you fired, that’s my point, that’s why what you’re saying is wrong:

Anonymous K replied with this 2 weeks ago, 1 minute later, 1 day after the original post[^] [v] #1,432,062

@1,432,060 (Darkness)
The meaning here is if you can automate 20% of labor after firing are accounted for.

Obviously you can't count productivity increases for workers who are layed off.

Anonymous K double-posted this 2 weeks ago, 1 minute later, 1 day after the original post[^] [v] #1,432,063

@1,432,061 (Darkness)
You keep hammering home a point we both already understand.

The 20% automation to reduce the workforce by 20% would need to be 20% of the labor is replaced by machines.

Saying "before firings" would imply boosting productivity by 20% and then, next, firing people. That wouldn't work because you would fire them after making that 20% and thus have lower than 100% productivity after.

Darkness replied with this 2 weeks ago, 2 minutes later, 1 day after the original post[^] [v] #1,432,064

@1,432,062 (K)
If you have 100 employees, they make $100,000 in total in a day, you increase their productivity by 20% with AI, now they’re making $120,000. If you then fire 20% of the employees, there are 80 employees. Combined these 80 employees make $96,000 in a day. That man’s that you have to account for $24,000. $24,000 is not 20% of $96,000 which is the amount the 80 workers are producing in total in an day after firing.

(Edited 19 seconds later.)

Anonymous K replied with this 2 weeks ago, 3 minutes later, 1 day after the original post[^] [v] #1,432,067

@previous (Darkness)
Which is why you account for the productivity increase after the firings, not before.

If you account for it before, and then fire, you are left with less than 100% of original productivity.

You keep explaining reciprocal inversion, but that's pointless because I'm already aware of it.

The original point I made was that AI can allow companies to reduce their workforce, even if the AI cannot do all tasks.

Darkness replied with this 2 weeks ago, 1 minute later, 1 day after the original post[^] [v] #1,432,069

@1,431,991 (H)

> My original point is what I stand by: If you can automate 20% of your labor (after firings, obviously) then you can reduce your labor force by 20%.

So what you’re saying is you fire 20% of the people, you automate 20% of the labor, then you fire 20% of the people?

That doesn’t make sense.

Darkness double-posted this 2 weeks ago, 1 minute later, 1 day after the original post[^] [v] #1,432,071

After firings you have 80 people, so to automate 20% of the labor after firings, that’s doing the work of 16 people. And you can’t reduce your labor force by 20% after that, because you just did that already, that would be doing it twice.

Anonymous K replied with this 2 weeks ago, 4 minutes later, 1 day after the original post[^] [v] #1,432,074

@1,432,069 (Darkness)
No, I'm saying that before you do anything, pre-firing you need a plan. At that point you account for how much productivity changes and make sure you account for how it will be acfter after people are fired as part of that plan.

If you have 5 employees, and can increase productivity by 25% then you can account for what that looks like after layoffs and conclude that 1/5th can be fired.

I'm not arguing the point, which is why I said I'm aware of reciprocal inversion. I never said your math was wrong.

My actual claim is that AI can reduce labor requirements for the same output, even if it can't automate all parts of a job.

They just need to account for how that productivity will be affted after the firings as part of their business plan.

Darkness replied with this 2 weeks ago, 1 minute later, 1 day after the original post[^] [v] #1,432,076

I think what you’re trying to say is if you can automate 20% of the labor before firings, then you can remove 20% of the employees. But that’s not true, you can only fire 20% of the employees if you automate 25% of the labor before firing if you assume that the AI has to be prompted by humans. You can only automate 20% of the labor and then fire 20% of the workers if the AI doesn’t need human intervention. Which is an important distinction, because those two scenarios have very different properties mathematically.

Anonymous K replied with this 2 weeks ago, 49 seconds later, 1 day after the original post[^] [v] #1,432,077

@previous (Darkness)
I just explained that's not my position.

Darkness replied with this 2 weeks ago, 1 minute later, 1 day after the original post[^] [v] #1,432,079

@previous (K)

This is your original position:

@1,431,918 (H)

> It doesn't need to replace the whole employee, it just needs to automate some of their work.
>
> Automate 20% of the work your scientists do and you can lay of 20% of them while maintaining output.

So you were assuming that AI doesn’t need to be prompted by humans.

Anonymous K replied with this 2 weeks ago, 1 minute later, 1 day after the original post[^] [v] #1,432,080

@previous (Darkness)
No, I just didn't specify whether that 20% was before or after accounting for firing because I assumed you had some common sense.

What I wrote was technically vague, but not wrong.

Darkness replied with this 2 weeks ago, 1 minute later, 1 day after the original post[^] [v] #1,432,081

Because essentially in what you originally said, if AI can do 20% of the work, you can lay off 20% of the employees, which implies that the employees productivity isn’t being boosted by AI, the AI is acting completely independently on its own, and can do the work of 20% of the employees regardless of whether the employees are there or not. Which is different from if AI requires humans to prompt it. It’s the difference between AI being its own independent being and AI being a tool.

Anonymous K replied with this 2 weeks ago, 1 minute later, 1 day after the original post[^] [v] #1,432,082

@previous (Darkness)
I never said the productivity increase would be per employee, nor did I say it would be a 20% increase before you fire people.

You made an assumption, I never said what you think I said. That's called a strawman, and you can easily avoid it by asking clarifying questions- but instead you explained multiplicative inversion many times.

Darkness replied with this 2 weeks ago, 12 seconds later, 1 day after the original post[^] [v] #1,432,083

@1,432,080 (K)

> No, I just didn't specify whether that 20% was before or after accounting for firing because I assumed you had some common sense.
>
> What I wrote was technically vague, but not wrong.

You later clarified:

@1,431,991 (H)

> My original point is what I stand by: If you can automate 20% of your labor (after firings, obviously) then you can reduce your labor force by 20%.

After firing 20% of your workers you only have 80% of the workers you started with, so 20% of 80% is 16% of your original workforce before firing, so it would make more sense if you said before not after.

Anonymous K replied with this 2 weeks ago, 2 minutes later, 1 day after the original post[^] [v] #1,432,084

@previous (Darkness)
I said "automate 20% after firings" in that post.

If the automation is after the firings, then you lose 20% of the workforce first, and then automate 20% after the firings. Note I never said it would be 20% increase per employee, you made that assumption yourself.

Darkness replied with this 2 weeks ago, 1 minute later, 1 day after the original post[^] [v] #1,432,085

@previous (K)
20% of what? After the firings, you only have 80% of the workforce you started with. The only thing you can possibly mean is 20% of the work done before the firings.

Darkness double-posted this 2 weeks ago, 2 minutes later, 1 day after the original post[^] [v] #1,432,086

If you’re talking about after firing 20% of people, then you would need to automate 25% of the work done, and then it would add up to the same amount of work as before the firings.

Anonymous K replied with this 2 weeks ago, 1 minute later, 1 day after the original post[^] [v] #1,432,087

@1,432,085 (Darkness)
> 20% of what? After the firings, you only have 80% of the workforce you started with.

You automate the 20% of labor you just fired. Whoch would mean a 25% per-employee increase.

> The only thing you can possibly mean is 20% of the work done before the firings.

It's odd you would say that, and bold the word before when I literally said "after firings".

I was making a joke when I asked if you thought after firings meant before firings, but apparently you really do think you can just infer the exact opposite of what I said.

Darkness replied with this 2 weeks ago, 56 seconds later, 1 day after the original post[^] [v] #1,432,088

Because 25% of 80% is 20%. If you automate 25% of the work the 80% of workers who are left over after firing can do, and then all those 80% of workers left over work as hard as they were working before the layoffs, plus that extra 25% productivity boost per person, that adds up to 100% of the productivity from beforehand.

Darkness double-posted this 2 weeks ago, 1 minute later, 1 day after the original post[^] [v] #1,432,090

@1,432,087 (K)

> > 20% of what? After the firings, you only have 80% of the workforce you started with.
>
> You automate the 20% of labor you just fired. Whoch would mean a 25% per-employee increase.
>
> > The only thing you can possibly mean is 20% of the work done before the firings.
>
> It's odd you would say that, and bold the word before when I literally said "after firings".


The reason why I said that is because what you’re saying doesn’t work mathematically and you said you were being vague so I should interpret you charitably and that’s the only way to interpret what you said in a way that works mathematically.

Anonymous K replied with this 2 weeks ago, 1 minute later, 1 day after the original post[^] [v] #1,432,091

@1,432,086 (Darkness)
You would need to automate 25% of the per-employee work, or the 20% of the total original work.

I didn't specify I was talking about total original work because it should be obvious, but at worst I've been vague. I never said I was talking about per-employee, and I didn't say before the firings (I literally said after).

Once again, you are arguing against something I never said, and you could avoid the strawman accusation by asking clarifying questions. You won't ever do that, no matter how many times I tell you how to avoid strawmanning, because you rely on the vagueness to argue against something dumb and make yourself feel smart. Asking those questions would eliminate that opportunity.

Anonymous K double-posted this 2 weeks ago, 1 minute later, 1 day after the original post[^] [v] #1,432,092

@1,432,090 (Darkness)
It's not my job to take accountability for your dumb assumptions.

If you want to really get me on something like this, ask if I mean "per employee, beforehand" but you won't because you rely on the vagueness to feel smart and argue against aomething dumb I never said.

Same in every thread. I've pointed out how you can avoid the accusation in many threads now.

Darkness replied with this 2 weeks ago, 5 seconds later, 1 day after the original post[^] [v] #1,432,093

@1,432,091 (K)

> You would need to automate 25% of the per-employee work, or the 20% of the total original work.

Yes, 20% of the original work is 20% of the work done before firing, not after firing.

Anonymous K replied with this 2 weeks ago, 2 minutes later, 1 day after the original post[^] [v] #1,432,094

@previous (Darkness)
Only if we are talking about total work (not per-employee) beforehand.

I never said it would be 20% per-employee before firing, but as usual you don't care if I actually said it.

Darkness replied with this 2 weeks ago, 1 second later, 1 day after the original post[^] [v] #1,432,095

Now do you understand the point I’m making about how there’s a difference between AI boosting the productivity of workers and AI working on its own? If AI works on its own, you can replace workers much more easily. If AI boosts the productivity of workers, the percentage of employees you can replace grows more slowly than the percentage increase in productivity per employee needed to maintain the same total productivity does.

Anonymous K replied with this 2 weeks ago, 2 minutes later, 1 day after the original post[^] [v] #1,432,097

@previous (Darkness)
I never said otherwise. I have no idea why you are arguing over and over and over for something I never disagreed about.

Darkness replied with this 2 weeks ago, 4 minutes later, 1 day after the original post[^] [v] #1,432,098

And do you understand my point that if you assume humans need to prompt chatbots (aka chatbots increase the productivity of employees), this is the function that determines what percentage increase in productivity per person AI has to achieve (y%) given what percentage of the workforce you want to fire (x%):

y = 100/(1-x/100)-100

And this is the inverse of that function:

x = 100y / (y + 100)

the limit of x as y approaches infinity is 100, so the function x never reaches 100, meaning if you need humans to prompt AI, you will experience diminishing returns in the number of people you can fire as AI increases in capability and you will never have a scenario where you need 0 employees.

Anonymous K replied with this 2 weeks ago, 1 minute later, 1 day after the original post[^] [v] #1,432,099

@previous (Darkness)
You're just going to keep pretending I disagree with you then, no matter how many times I tell you I already know this?

Darkness replied with this 2 weeks ago, 1 minute later, 1 day after the original post[^] [v] #1,432,100

@previous (K)
You clearly didn’t understand the point I made.

@1,431,949 (H)

> > people use AI, AI isn’t self directed because it’s not actually intelligent.
>
> What does that change? It's just like any other automation tool that people use. If a power tool makes a construction worker 20% more efficient, and it isn't self-directed, but must be held and operated by a construction worker, that tool can still allow a company to lay off 20% of their staff to maintain the same level of production. Why would AI be any different?

Darkness double-posted this 2 weeks ago, 1 minute later, 1 day after the original post[^] [v] #1,432,101

I posted the entire explanation of why the difference between these two cases matters based on whether AI is self directed or not and you responded with "What does that change?"

You also went on a whiny insecure racist rant because you realized I’m smarter than you are.

@1,432,054 (K)

> Your reading comprehension is abysmal. This is why they had to get rid of the SATs, only blacks could be fooled into thinking 6 figure debt is smart for a degree, but blacks can't pass literacy tests. Even blacks fought to say literacy tests are racist.

Anonymous K replied with this 2 weeks ago, 2 minutes later, 1 day after the original post[^] [v] #1,432,103

@previous (Darkness)

> I posted the entire explanation of why the difference between these two cases matters based on whether AI is self directed or not and you responded with "What does that change?"

Because my point was that AI can replace jobs without replacing everything about a job. It's irrelevant to the matter of whether someone needs to prompt it.

> You also went on a whiny insecure racist rant because you realized I’m smarter than you are.

You made a dozen posts trying to convince me of something that I said over and over I agree with you on.

> > Your reading comprehension is abysmal. This is why they had to get rid of the SATs, only blacks could be fooled into thinking 6 figure debt is smart for a degree, but blacks can't pass literacy tests. Even blacks fought to say literacy tests are racist.

Yes, well you proved that here, didn't you? You tried multiple times to demonstrate something no one disagreed on.

Then I posted that, and you made many many more posts trying to convince me of it still, even though I kept saying I already knew and agreed.

Darkness replied with this 2 weeks ago, 59 seconds later, 1 day after the original post[^] [v] #1,432,104

@previous (K)
The only thing I proved is you don’t know how to do 4th grade math.

Anonymous K replied with this 2 weeks ago, 2 minutes later, 1 day after the original post[^] [v] #1,432,105

@previous (Darkness)

I agreed with the 4th grade math, and even used the exact name of what you were talking about (multiplicative inversion) and you just kept pretending I disagreed.

What's the point of all this? Do you think that if you ignore every post where I say that "I know, I don't disagree with your math" that someone will open up the thread and just assume there was a reason for you doing that?

Darkness replied with this 2 weeks ago, 58 seconds later, 1 day after the original post[^] [v] #1,432,106

@previous (K)
No you argued with me about it. You wouldn’t have done that and then went on a racist rant about my intelligence if you understood math.

Anonymous K replied with this 2 weeks ago, 36 seconds later, 1 day after the original post[^] [v] #1,432,107

@1,432,082 (K)
Here's the exact post where I said this was called multiplicative inversion, btw.

Anonymous K double-posted this 2 weeks ago, 34 seconds later, 1 day after the original post[^] [v] #1,432,108

@1,432,106 (Darkness)

> No you argued with me about it.

No, I didn't. Find one post where I said your math was wrong. You won't though.

Darkness replied with this 2 weeks ago, 1 minute later, 1 day after the original post[^] [v] #1,432,109

@previous (K)
@1,431,991 (H)

> Great, you also made mistakes in the thread. Like thinking someone needs to prompt that AI somehow changes anything about this.

Isn’t that what my mathematical argument was about?

Darkness double-posted this 2 weeks ago, 44 seconds later, 1 day after the original post[^] [v] #1,432,110

If you thought my argument was correct, why did you say that?

Darkness triple-posted this 2 weeks ago, 1 minute later, 1 day after the original post[^] [v] #1,432,111

And why would you insult my intelligence on the basis of race if you think I’m making a correct argument?

Darkness quadruple-posted this 2 weeks ago, 49 seconds later, 1 day after the original post[^] [v] #1,432,112

You know, the most embarrassing way to lose an argument has to be pretending you never disagreed with me.

Anonymous K replied with this 2 weeks ago, 3 minutes later, 1 day after the original post[^] [v] #1,432,113

@1,432,109 (Darkness)
When I have said this whole time that it has to account for the productivity after the firings, it'a obviously not what I'm saying.

@1,432,110 (Darkness)

> If you thought my argument was correct, why did you say that?

Because my point was that you don't need to automate an entire job to have job losses. Even if parts are not able to be automated you can still reduce the labor and reassign people to do more of the work that can't be automated.

@1,432,111 (Darkness)
Because you tried explaining multiplicative inversion a dozen times, ignoring the fact that I said I didn't disagree with the math after each one.

Explaining it once isn't stupid, maybe there was a miscommunication.

Explaining it a dozen times when I keep saying I know already is psychotic behavior that makes it look like you're incapable of reading the multiple posts I make where I explain the math isn't the problem.

Darkness joined in and replied with this 2 weeks ago, 2 minutes later, 1 day after the original post[^] [v] #1,432,114

@previous (K)
Even though you’re pretending to agree with me now, I’m not convinced that you understand my argument about limits given how much time you spent arguing with me about percentages.

(Edited 15 seconds later.)

Anonymous K replied with this 2 weeks ago, 37 seconds later, 1 day after the original post[^] [v] #1,432,115

@1,432,112 (Darkness)
If someone says "I don't disagree" a dozen times, and each time you act like they do disagree and you reexplain the thing that they keep saying they already understand you don't think that's embarassing?

Most people would just stop after once or twice and tey to figure out what the actual contention is rather than acting like the other person is disagreeing when they keep saying the exact opposite.

How many times do you need to hear it? Even if you really thought I disagreed at first (and I didn't) why not think "oh, he said the math is right, I don't need to explain this anymore"?

Darkness replied with this 2 weeks ago, 16 seconds later, 1 day after the original post[^] [v] #1,432,116

And the racist rant was immature. I’m clearly a lot smarter than you are.

Anonymous K replied with this 2 weeks ago, 1 minute later, 1 day after the original post[^] [v] #1,432,117

@1,432,114 (Darkness)
I never disagreed with the math.

Very early on I said you were right about the math.

(Edited 24 seconds later.)

Darkness replied with this 2 weeks ago, 1 minute later, 1 day after the original post[^] [v] #1,432,118

@previous (K)
And yet you kept saying things to contradict my argument anyway because you didn’t understand it.

Anonymous K replied with this 2 weeks ago, 21 seconds later, 1 day after the original post[^] [v] #1,432,119

@1,432,116 (Darkness)
You just made over a dozen posts acting like I disagreed with you when I said the math was right from the start.

That isn't smart behavior.

Smart people can give direct quotes, and move on when someone agrees with them. They don't need to explain something a dozen times when it's already agreed by everyone in the conversation.

Darkness replied with this 2 weeks ago, 38 seconds later, 1 day after the original post[^] [v] #1,432,120

@previous (K)
Smart people can think. You can’t think. The only thing you can do is copy. You’re not smart.

Darkness double-posted this 2 weeks ago, 50 seconds later, 1 day after the original post[^] [v] #1,432,121

If you could think you would realize how idiotic it sounds that you’re accusing me of being stupid because I’m correct. I can’t be stupid for being right.

Anonymous K replied with this 2 weeks ago, 9 seconds later, 1 day after the original post[^] [v] #1,432,122

@1,432,118 (Darkness)
@1,431,969 (H)

> I said if you automate 20% of the work, you can replace 20% of your workforce.
>
> If you have 5 employees, and want to fire one, yes, you will need to increase productivity *per employee* by 25% to automate 20% of the work after the firing.
>
> The original point I made still stands. 20% automation = 20% reduction in labor force.

This post was three hours ago.

If you read that, and thought I disagreed with you, you are conpletely illiterate and cannot comprehend the things you are reading.

Darkness replied with this 2 weeks ago, 54 seconds later, 1 day after the original post[^] [v] #1,432,123

@previous (K)
Yeah, and I kept arguing with you, because you made a statement that is incorrect that directly contradicts the point I made.

Anonymous K replied with this 2 weeks ago, 2 minutes later, 1 day after the original post[^] [v] #1,432,124

@1,432,122 (K)
Actually its crazy that I spelled out the exact math of increasing per-employee productivity by 25% to get 20% total productivity increase after the firing- and you spent three hours arguing FOR the thing I literally just said.

Smart people do not insist they are being disagreed for three hours with when the other person literally lays out the exact argument clearly and precisely first.

This is actually one of the stupidest debates you've had, because no one denied it, and I said it multiple times.

You are so desperate to seem smart, but anyone opening this thread can see why racism is so prevalent avainst your kind.

I'm cutting my losses, it shouldn't take you 3 hours to understand that someone who lays it out that clearly is not disagreeing. You may actually qualify for disability at this point.

Anonymous K double-posted this 2 weeks ago, 49 seconds later, 1 day after the original post[^] [v] #1,432,125

@1,432,123 (Darkness)
I literally laid out the exact point that you need 25% increase per-employee for 4 employees to do the work of 5.

You are actually a retarded nigger, bye.

Darkness replied with this 2 weeks ago, 16 seconds later, 1 day after the original post[^] [v] #1,432,126

@1,432,124 (K)
Racism is prevalent against African Americans because you lost a debate about 4th grade math on the internet?

Darkness double-posted this 2 weeks ago, 27 seconds later, 1 day after the original post[^] [v] #1,432,127

@1,432,125 (K)

> You are actually a retarded nigger, bye.

I’m just imagining you crying while typing this lol

Darkness triple-posted this 2 weeks ago, 3 minutes later, 1 day after the original post[^] [v] #1,432,128

And hey, I might be black, but don’t forget, I’m also Irish. If you want me to quit, I don’t know how tf to do it. I couldn’t quit even if I wanted to.

Anonymous M joined in and replied with this 2 weeks ago, 37 seconds later, 1 day after the original post[^] [v] #1,432,129

@1,432,126 (Darkness)
he said you need 25 percent tho?

Anonymous K replied with this 2 weeks ago, 41 seconds later, 1 day after the original post[^] [v] #1,432,130

@previous (M)
He's either on meth and can't follow the conversation or he's trying to make blacks look bad.

Darkness replied with this 2 weeks ago, 36 seconds later, 1 day after the original post[^] [v] #1,432,131

@previous (K)
How come I can be calm and you’re getting emotional?

Darkness double-posted this 2 weeks ago, 1 minute later, 1 day after the original post[^] [v] #1,432,132

I never insult anyone, especially not you. I’ve always treated you with respect, and it shameful that this is how you return the favor.

Darkness triple-posted this 2 weeks ago, 28 minutes later, 1 day after the original post[^] [v] #1,432,137

Now that I have won the debate

Who wants to hear some fresh pizza hacks from Danny fresh

BonelessPizzaGang quadruple-posted this 2 weeks ago, 23 minutes later, 1 day after the original post[^] [v] #1,432,140

White crybabies after losing 4th grade math debates be like:

"I hate these monkey niggers! 😭😭😭 Why do they have big black dicks? Why do these big black gorilla niggers with their big black clocks have to live here in the west? Why did we kidnap them and bring them here? I hate women but I want a girlfriend! Why do white women hate their own race! 😭😭😭 I’m gonna move to China and get a smoking hot Chinese wife and then you’ll all be sorry! 我爱中国和我爱习近平。美国人是同性恋!"
https://youtu.be/006FqsXs2Do

(Edited 29 seconds later.)

:

Please familiarise yourself with the rules and markup syntax before posting.