5 ms·

I don't think we can say the AI Winter is here until these large funding rounds stop.

itake 8y ago
These funding rounds just mean we are using existing tools. The AI winter is regards to discovering new AI techniques, not re-applying existing discoveries to business.
- kinsomo 8y ago
  > These funding rounds just mean we are using existing tools. The AI winter is regards to discovering new AI techniques, not re-applying existing discoveries to business. There's so much hype around "AI" right now that I wonder if we're headed to a winter where you need very strong business justification to get funding to apply those existing tools. It seems like anyone with a problem to solve is spending millions to throw machine learning at it, just to see if it will stick. That's not sustainable.
pietroglyph 8y ago
I think that even if research just stopped (like in an AI winter), there's still a lot of applications for the tech we have now. I suspect that's where most of this money will go.
seanmcdirmid 8y ago
I don’t think we can say that the AI winter is coming without these large funding rounds.
roymurdock 8y ago
From the author of the AI Winter post: "I don't think the winter depends on the amount that somebody makes today on AI, rather on how much people are expecting to make in the future. If these don't match, there will be a winter. My take is that there is a huge bet against the future. And if DL ends up bringing just as much profit as it does today, interest will die very, very quickly."
- alex_young 8y ago
  This possibility is interesting to consider from an economic perspective. Let's posit that DL does just that, bring in the same value (as a whole system) as non-DL. Inside such a system, there is a business advantage to implementation of DL, and there is a concentration of value around those with the largest dataset. This means that DL systems will still be very valuable in that they are a competitive advantage, but only so much as they are difficult to implement independently. The value differentiator then becomes the data itself. I think this points to a value proposition for data providers. GDPR and such seem to be quite correctly identifying and correcting for this risk.
exelius 8y ago
I don’t think there will be an AI winter frankly. A lot of the use cases that businesses want can be built with a combination of decision trees and some ML for classification. ML works incredibly well on even lightly-structured data (which includes most financial data streams). Agree it hasn’t lived up to the AI hype, but that doesn’t mean what it can deliver on isn’t useful.
gnode 8y ago
AI Winter was referring to how the potential of current AI methods is limited, and aren't going to naturally evolve into artificial general intelligence. However I don't think it matters in this case. There are still many untapped applications of the AI technologies that work today. Deploying recognition technologies at scale in China to strengthen the government's control over the populace is one of them.
- kinsomo 8y ago
  > AI Winter was referring to how the potential of current AI methods is limited Not exactly. "AI Winter" refers the historical periods that followed unsustainable hype that surrounds AI methods prior to those times. Skepticism of AI during the last winter was so strong that "AI" kinda became a dirty word and people invented euphemisms like "machine learning" to avoid the stigma. https://en.wikipedia.org/wiki/AI_winter https://en.wikipedia.org/wiki/AI_winter

These funding rounds just mean we are using existing tools. The AI winter is regards to discovering new AI techniques, not re-applying existing discoveries to business.

kinsomo 8y ago
> These funding rounds just mean we are using existing tools. The AI winter is regards to discovering new AI techniques, not re-applying existing discoveries to business. There's so much hype around "AI" right now that I wonder if we're headed to a winter where you need very strong business justification to get funding to apply those existing tools. It seems like anyone with a problem to solve is spending millions to throw machine learning at it, just to see if it will stick. That's not sustainable.

pietroglyph 8y ago

I think that even if research just stopped (like in an AI winter), there's still a lot of applications for the tech we have now. I suspect that's where most of this money will go.

seanmcdirmid 8y ago

I don’t think we can say that the AI winter is coming without these large funding rounds.

roymurdock 8y ago

From the author of the AI Winter post: "I don't think the winter depends on the amount that somebody makes today on AI, rather on how much people are expecting to make in the future. If these don't match, there will be a winter. My take is that there is a huge bet against the future. And if DL ends up bringing just as much profit as it does today, interest will die very, very quickly."

alex_young 8y ago
This possibility is interesting to consider from an economic perspective. Let's posit that DL does just that, bring in the same value (as a whole system) as non-DL. Inside such a system, there is a business advantage to implementation of DL, and there is a concentration of value around those with the largest dataset. This means that DL systems will still be very valuable in that they are a competitive advantage, but only so much as they are difficult to implement independently. The value differentiator then becomes the data itself. I think this points to a value proposition for data providers. GDPR and such seem to be quite correctly identifying and correcting for this risk.

exelius 8y ago

I don’t think there will be an AI winter frankly. A lot of the use cases that businesses want can be built with a combination of decision trees and some ML for classification. ML works incredibly well on even lightly-structured data (which includes most financial data streams). Agree it hasn’t lived up to the AI hype, but that doesn’t mean what it can deliver on isn’t useful.

gnode 8y ago

AI Winter was referring to how the potential of current AI methods is limited, and aren't going to naturally evolve into artificial general intelligence. However I don't think it matters in this case. There are still many untapped applications of the AI technologies that work today. Deploying recognition technologies at scale in China to strengthen the government's control over the populace is one of them.

kinsomo 8y ago
> AI Winter was referring to how the potential of current AI methods is limited Not exactly. "AI Winter" refers the historical periods that followed unsustainable hype that surrounds AI methods prior to those times. Skepticism of AI during the last winter was so strong that "AI" kinda became a dirty word and people invented euphemisms like "machine learning" to avoid the stigma. https://en.wikipedia.org/wiki/AI_winter https://en.wikipedia.org/wiki/AI_winter