IBM Unveils New Watsonx, AI and Data Platform (reuters.com) 22
IBM on Tuesday launched watsonx, a new artificial intelligence and data platform to help companies integrate AI in their business. From a report: The new AI platform launch comes over a decade after IBM's software called Watson got attention for winning the game show Jeopardy. IBM at the time said Watson could "learn" and process human language. But Watson's high cost at the time made it a challenge for companies to use, according to Reuters reporting. Fast forward a decade, chatbot ChatGPT's overnight success is making AI adoption at companies a focus, and IBM is looking to grab new business. This time, the lower cost of implementing the large language AI models means the chances of success are high, IBM CEO Arvind Krishna told Reuters ahead of the company's annual Think conference.
"When something becomes 100 times cheaper, it really sets up an attraction that's very, very different," said Krishna. "The first barrier to create the model is high, but once you've done that, to adapt that model for a hundred or a thousand different tasks is very easy and can be done by a non-expert." Krishna said AI could reduce certain back office jobs at IBM in the coming years. "That doesn't mean the total employment decreases," he said about some media reports talking about IBM pausing hiring for thousands of jobs that AI could replace. "That gives the ability to plow a lot more investment into value-creating activities...We hired more people than were let go because we're hiring into areas where there is a lot more demand from our clients."
"When something becomes 100 times cheaper, it really sets up an attraction that's very, very different," said Krishna. "The first barrier to create the model is high, but once you've done that, to adapt that model for a hundred or a thousand different tasks is very easy and can be done by a non-expert." Krishna said AI could reduce certain back office jobs at IBM in the coming years. "That doesn't mean the total employment decreases," he said about some media reports talking about IBM pausing hiring for thousands of jobs that AI could replace. "That gives the ability to plow a lot more investment into value-creating activities...We hired more people than were let go because we're hiring into areas where there is a lot more demand from our clients."
Pronunciation (Score:5, Funny)
I am ready to accept my consulting fee now.
Re: (Score:2)
Maybe they didn't learn anything from Microsoft's attempt at a silent X.
If 80% of IBM's managers are replaced by AI... (Score:2, Interesting)
.... will any IBM employees that actually create value notice any difference? Other than those who find it an improvement.
Re: (Score:2)
Maybe the AI will waste less of their time in meetings because an AI doesn't have an ego that needs constant feeding?
Let's not kid ourselves (Score:5, Interesting)
The reason Watson didn't catch on wasn't because of Watson's high cost, it was because of IBM Services high cost. Their CEO is talking about AI becoming 100 times cheaper, but I would venture to guess that the cost to IBM customers isn't even half. I saw it first hand when I was in Global Services and Armonk started moving to the "global factory" model for consultants. (also, isn't it weird that IBM's CEO thinks employees need to be in the office, but thinks a consultant 8000 miles away can support a client? But I digress.) Global Services Clients caught on really fast when IBM press releases talked about how much money they were saving using Indian labor at 1/10th the cost of North America employees and they saw 10% or more of the onsite IBMers laid off each year. The minute clients asked why the savings weren't being passed on to them and IBM just shrugged, that's usually when the decision was made not to renew the contract.
Re:Let's not kid ourselves (Score:4, Funny)
Pretty much this. IBM has a tendency to blow their chances by charging about 95% for a service compared to a full fledged human thinking that clients will go nuts at a tiny cost savings, with the 'IBM' brand to bolster the value to nearly what they would have paid anyway. That's if you are *lucky*, often they would just charge even more for the cheap service than you could have just done it yourself.
So even if IBM can hypothetically offer a service at 5% of the cost otherwise, they will not settle for anything less than 90%.
Re: (Score:2)
The reason Watson didn't catch on wasn't because of Watson's high cost, it was because of IBM Services high cost.
Watson was also quite shitty. I had the opportunity to ask it some test questions when its use as a medical adviser was touted as one of its most advanced applications. The responses were just appalling, and - unlike today's LLM responses - often did not even address the core of the question asked. Did not even reach the level where it would have made sense to check responses for inaccuracies.
In 2011, Watson amazed everyone (Score:2, Interesting)
In 1997, IBM's Deep Blue beat Gary Kasparov at Chess and while we knew it was inevitable, we were blown away. But we felt safe because it cost $100 million, wasn't AGI. And it still couldn't play GO.
In 2011, IBM's Watson won a Jeopardy tournament. Again we were all blown away, but we felt safe because it cost $1 billion and it still wasn't AGI.
In 2017, Google's DeepMind AlphaGo defeated Ke Jie at Go. But we felt safe because it wasn't AGI, it just played GO.
In 2022, ChatGPT can do everything Watson coul
Re:In 2011, Watson amazed everyone (Score:4, Interesting)
Note that the gap between 2011 and 2022 was not so huge. Note that it did not cost a Billion dollars. IBM purported to spend a billion in general on the Watson brand by 2014, including all sorts of marketing and acquisitions and such. Also, beware of 'IBM accounting', bragging about investing a Billion when there's lots of room for fanagling. They may have decided, for example, that X% of the POWER7 costs should be attributed to 'Watson', even though that money was going to be spent with or without Watson. The actual PR stunt was done on a pretty vanilla 90 node POWER cluster that wouldn't have even come close to the Top500 of the time, not even half of the slowest Top500 system.
Watson on Jeopardy was not some impossibly expensive moonshot, it was a solid demo of a tech that ultimately they couldn't find practical application for, at least not at the margins IBM would demand of it...
Re: (Score:2)
I agree that LLM poetry does not understand meter.
Multimodal:
I think OpenAI's current beta generation of GPT-4 is multimodal in that it takes images as input. I am excited for that. I'm unclear how they are doing it: is it another AI/algorithm that then feeds data into the LLM? In another thread, there was talk about LLMs for the military where I argued that an LLM is not useful for strategic planning. In that, I posited that one could use the LLM as the front-end for some other kind of AI that is capab
Re: (Score:1)
Re: (Score:2)
In short, "Hey, we see OpenAI being fawned over like crazy, we'd like to remind everyone that we tried (and failed) at this already, and you should be investing in us because if OpenAI has done anything, it's to prove that IBM was right?"
How about RPGs? (Score:2)
Will we finally get NPCs that can talk back and be moody and grumpy?
Why would I give IBM money? (Score:2)
Capability query (Score:2)
Reminds me of Dot Com Bust times (Score:1)
This time around experienced people will be made to train AI and then will laid off, and AI will do the consulting/coding.