Neither is saying that it's using your processing power in a way that is some situations is quite similar to how Google uses people to teach their AIs image recognition with "training CAPTCHAs".
Cortana still cant answer "what time is sunset tonight", I'm not holding my breath for an "engaging and magical" experience from the gang in Redmond anytime soon.
I guess because "it just works" and integrates nicely with their usual environment. Don't know if TensorFlow provides the same flexibility and ease of use as MS promises here.
It would be really helpful for the Win environment if people used this to make better software, leveraging the many core CPUs and powerful GPUs available.
It just works? So you've developed some products with it already I see. Very good.
It would be really helpful for Win ev. ppl if they used this to leverage CPU/GPU cores... um... Cuda/OpenCL etc have been around a long time for that. An AI API isn't going to magically make them start making better use of parallelism. If anything this is going to encourage lazy programmers to burn cycles pointlessly using AI for things that better optimisation algorithms would have done quicker, more accurately, and with way way way less power.
What it will do a brilliant job of is monetizing the Windows subscriber base by targeting them with app/product recommendations and referrals, which is the real reason Redmond is in such a rush to land this on the desktop. Secondary to that is improving on-screen keyboard predictions and voice recognition, possibly improve touchscreen accuracy too by recognizing failure patterns and ignoring them. Tertiary to that is whatever developers want to do with it... more advertising, user monitoring, and perhaps a small amount of actual AI towards optimising UI's, resource usage, and gaming experiences.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
9 Comments
Back to Article
shabby - Wednesday, March 7, 2018 - link
Ai this, ai that, this word is already losing its meaning.Stochastic - Wednesday, March 7, 2018 - link
Pattern recognizer isn't as marketable.close - Thursday, March 8, 2018 - link
Neither is saying that it's using your processing power in a way that is some situations is quite similar to how Google uses people to teach their AIs image recognition with "training CAPTCHAs".Holliday75 - Wednesday, March 7, 2018 - link
Pre-programmed intelligence.Gunbuster - Thursday, March 8, 2018 - link
Cortana still cant answer "what time is sunset tonight", I'm not holding my breath for an "engaging and magical" experience from the gang in Redmond anytime soon.peevee - Thursday, March 8, 2018 - link
And why would anybody limit themselves to MS, Windows 10, DX12 and Win ML when you can pick up a multi-platform API like TensorFlow?MrSpadge - Thursday, March 8, 2018 - link
I guess because "it just works" and integrates nicely with their usual environment. Don't know if TensorFlow provides the same flexibility and ease of use as MS promises here.MrSpadge - Thursday, March 8, 2018 - link
It would be really helpful for the Win environment if people used this to make better software, leveraging the many core CPUs and powerful GPUs available.linuxgeex - Thursday, March 22, 2018 - link
Armchair response much?It just works? So you've developed some products with it already I see. Very good.
It would be really helpful for Win ev. ppl if they used this to leverage CPU/GPU cores... um... Cuda/OpenCL etc have been around a long time for that. An AI API isn't going to magically make them start making better use of parallelism. If anything this is going to encourage lazy programmers to burn cycles pointlessly using AI for things that better optimisation algorithms would have done quicker, more accurately, and with way way way less power.
What it will do a brilliant job of is monetizing the Windows subscriber base by targeting them with app/product recommendations and referrals, which is the real reason Redmond is in such a rush to land this on the desktop. Secondary to that is improving on-screen keyboard predictions and voice recognition, possibly improve touchscreen accuracy too by recognizing failure patterns and ignoring them. Tertiary to that is whatever developers want to do with it... more advertising, user monitoring, and perhaps a small amount of actual AI towards optimising UI's, resource usage, and gaming experiences.