AI's Costly Buildup Could Make Early Products a Hard Sell 24
Microsoft, Google and others experiment with how to produce, market and charge for new tools. From a report: Microsoft has lost money on one of its first generative AI products, said a person with knowledge of the figures. It and Google are now launching AI-backed upgrades to their software with higher price tags. Zoom has tried to mitigate costs by sometimes using a simpler AI it developed in-house. Adobe and others are putting caps on monthly usage and charging based on consumption. "A lot of the customers I've talked to are unhappy about the cost that they are seeing for running some of these models," said Adam Selipsky, the chief executive of Amazon.com's cloud division, Amazon Web Services, speaking of the industry broadly. It will take time for companies and consumers to understand how they want to use AI and what they are willing to pay for it, said Chris Young, Microsoft's head of corporate strategy.
"We're clearly at a place where now we've got to translate the excitement and the interest level into true adoption," he said. Building and training AI products can take years and hundreds of millions of dollars, more than with other types of software. AI often doesn't have the economies of scale of standard software because it can require intense new calculations for each query. The more customers use the products, the more expensive it is to cover the infrastructure bills. These running costs expose companies charging flat fees for AI to potential losses.
Microsoft used AI from its partner OpenAI to launch GitHub Copilot, a service that helps programmers create, fix and translate code. It has been popular with coders -- more than 1.5 million people have used it and it is helping build nearly half of Copilot users' code -- because it slashes the time and effort needed to program. It has also been a money loser because it is so expensive to run. Individuals pay $10 a month for the AI assistant. In the first few months of this year, the company was losing on average more than $20 a month per user, according to a person familiar with the figures, who said some users were costing the company as much as $80 a month.
"We're clearly at a place where now we've got to translate the excitement and the interest level into true adoption," he said. Building and training AI products can take years and hundreds of millions of dollars, more than with other types of software. AI often doesn't have the economies of scale of standard software because it can require intense new calculations for each query. The more customers use the products, the more expensive it is to cover the infrastructure bills. These running costs expose companies charging flat fees for AI to potential losses.
Microsoft used AI from its partner OpenAI to launch GitHub Copilot, a service that helps programmers create, fix and translate code. It has been popular with coders -- more than 1.5 million people have used it and it is helping build nearly half of Copilot users' code -- because it slashes the time and effort needed to program. It has also been a money loser because it is so expensive to run. Individuals pay $10 a month for the AI assistant. In the first few months of this year, the company was losing on average more than $20 a month per user, according to a person familiar with the figures, who said some users were costing the company as much as $80 a month.
Not surprised. (Score:2)
So far, "AI" (meaning: large pattern-recognition and replication software) has been laughable, and not yet ready to replace humans. But it's getting there. In the future, though, unless the rate of progress changes, it will replace a lot of humans, and therefore could be quite profitable.
So, it's not surprising that, to start with, AI is costing more than it's earning.
Re: (Score:2)
I agree with your assessment of current "AI" being laughable pattern recognition.
Re:Not surprised. (Score:5, Interesting)
Not surprised.
So far, "AI" (meaning: large pattern-recognition and replication software) has been laughable, and not yet ready to replace humans. But it's getting there. In the future, though, unless the rate of progress changes, it will replace a lot of humans, and therefore could be quite profitable.
So, it's not surprising that, to start with, AI is costing more than it's earning.
Most of the big "AI" pushers right now are in hard R&D mode. Something that with most forms of software wouldn't be public facing. But, since our current version of "AI" is mostly about data-aggregation, the simplest way to get massive piles of data is to sucker plebes into providing it. Voila, we are no longer beta testers, we are a defacto R&D department. So, sure, people may be losing money in this particular moment while developing AI. Once it gets to the point where it's producing repeatable, reputable, reliable results, that money flow will change directions.
I'm still not sure why so many of us have been so eager to train our eventual replacements without any compensation. In fact, large numbers of us are now paying for the privilege. Mostly under the adage, "Gotta keep up with the times or you'll be replaced faster." And that's the best we can hope for. Be replaced quick, or be replaced soon. What a great time to be alive.
Re: Not surprised. (Score:2)
"Once it gets to the point where it's producing repeatable, reputable, reliable results, that money flow will change directions."
I'm not sure that's true, or at least, that it will come that soon. For the foreseeable future there will still be pressure to continually improve, due to competition.
Re: (Score:2)
amazing.
that no one has thought about using the open a i tool for topics.
like applied to veterans benefits
It's hard to make cents with AI (Score:3, Funny)
If automation is an added cost. (Score:1)
Since the cost of automation is just an increase in expenses.
If automation is cutting operations and production expenses then the pricing is easy.
And automation is a much easier to market and sell since lowering expenses adds to the bottom line.
Stop loss leading (Score:2)
Code completion and prompted inpainting are two of the applications with a very clear value proposition, way more than 10$ a month. Give one month of high use for 10 bucks and after that charge enough to make profits. Even for outsourced professionals in low wage nations, 10$ a month is silly little money for a significant productivity boost.
Of course I don't think code completion and prompted inpainting are exactly a trillion dollar market, there will have to be some correction on the larger market.
Re: (Score:1)
Mistral & LLaMA (Score:3)
If it can replace employees (Score:4, Insightful)
Re: (Score:1)
IBM PCs were $10-$15k each on launch
Probably important to note that's the price adjusted for inflation, not the actual price; the base price was $1565, but that didn't include much (not even a disk drive!). A more typical (and useful) configuration was in the $3000 - $4000 range, adjust those for inflation and you get closer to the $10-$15k range.
Re: (Score:2)
To be fair folks were buying $18k mini computers *without* adjusting for inflation to replace those junior accountants. The IBM PC at $3k would've seemed like a bargain too good to be true
Also I googled "mini computer" to make sure my memory on pricing wasn't too far off and the only thing that comes up by default are those little bric PC
Still cheaper than humans for now (Score:3)
Re: (Score:2)
AI IQ (Score:2)
AI doesn't need a 200 IQ to be useful. It only needs 70 IQ, maybe 50?
Except, current AIs aren't even close, and perpetually 10 years away...
Managers do need an IQ of +120 (Score:2)
Managers with an IQ of 70 will just throw all their money at the AI hype as some GPT chatbot told them that AI is the future.
So far they're expensive tools... (Score:2)
Re: (Score:2)
Even my company bought into the AI hype (Score:2)
an lost about $2 million in investments and ended up with a poor product.
Re: (Score:2)
Even I personally have bought into the AI hype. GitHub Copilot is totally worth the price to me, even though it's still pretty raw.
Of course, major new technology is expensive! (Score:2)
Every new technology costs a lot of money to develop. That's why startups raise money, because they know that at first, they will lose money, a lot of it. Why should AI be any different?
Charge per use, great idea (Score:2)