A few years back IBM was promoting their AI based API, Watson, everywhere. But it failed like anything. Now GitHub copilot seems promising and the industry is actively integrating it into their system. I didn't look into Watson deeply. I would like to know what is the difference in vision of that leadership that made Watson fail but the copilot succeeded (too early to say though)?
Sometimes the product is too far ahead of the market.
This is the (mostly) correct answer. For disclosure I worked closely with this group for a decade. People forget how early and far ahead IBM was in AI. Watson played jeopardy in 2011. That was probably 2 years ahead of the closest competitors at the time and completely ahead of the mainstream consciousness. People on the street struggled with what it meant. People in tech struggled with how to use (still going on now). I think the big failing from IBM happened at the leadership level and on the ground. Leadership was making really large claims. This culminated with Ginni R. Said Watson was going to cure cancer and spun up a business unit around it. On The ground, API calls was a seemingly intelligent business model but it left a lot of work to the customer. 2 problems with this. IBM was losing the cloud platform war at this time. And AI was slow to be embedded into solutions in anything like a seamless way. This left a lot of gaps in IBM’s AI strategy and competitors have filled those gaps. IBM also let the marketing run away from reality. In short, it’s not as exciting of a story as you may hope, there isn’t some rebellion or crazy upheaval behind it. IBM just moved first and misstepped by not providing a cloud platform that anyone would use to build on, made it a few steps too complicated to use the services, and let marketing drive the narrative too much. IBM’s still recovering from this, but is a great company with the right leadership in place. They’ll be just fine.
Marissa Mayer and Ginni are examples of what a leader should not be like.
This is Bing's response to "Why Watson failed". But these can happen to any AI model and could have been improved over time.
This is a great description fromchatGPT I would add 1. It was not properly monitized 2. It had no cloud deployment model 3. It was horrendously expensive to setup, and required a massive reconfig whenever your business or data change d in any meaningful way 3. It was definately not scalable, which is also true of chatgpt, but microsoft has basically free azure compute right now so they might be ok.
I’m not close to Watson or IBM but can relate to the decision making process at a shareholder driven company. Watson was a cash burner and I think IBM leaders found it tough to justify costs. Not every one gets a blank check like Amazon. More than market not being ready, I believe it has more to do with how the product was packaged. Unlike other AI tools, it was enterprise first and that was never moving to be an easy sell in the changing world.
Tech Industry
Yesterday
2127
The end of Backdoor Roth?!
India
Yesterday
927
Modi is a legend, will be remembered for centuries to come
Working Parents
Yesterday
872
What do you think is wrong with a kid who got rejected by 9 colleges?
Tech Industry
Yesterday
2699
Quitting this Slave life
Working Parents
12h
1004
Closed now - thank you all
Ask chatgpt