Google AI Chief Jeff Dean on Machine Learning Trends To Watch in 2020 (venturebeat.com) 15
In a wide-ranging interview with VentureBeat, Google AI chief Jeff Dean has discussed the company's early work on the use of ML to create semiconductors for machine learning, the impact of Google's BERT on conversational AI, and machine learning trends to watch in 2020. An excerpt from the interview where Dean talks about some of the trends one could expect to emerge, or milestones he thinks might be surpassed in 2020 in AI: I think we'll see much more multitask learning and multimodal learning, of sort of larger scales than has been previously tackled. I think that'll be pretty interesting. And I think there's going to be a continued trend to getting more interesting on-device models -- or sort of consumer devices, like phones or whatever -- to work more effectively. I think obviously AI-related principles-related work is going to be important. We're a big enough research organization that we actually have lots of different thrusts we're doing, so it's hard to call out just one. But I think in general [we'll be] progressing the state of the art, doing basic fundamental research to advance our capabilities in lots of important areas we're looking at, like NLP or language models or vision or multimodal things. But also then collaborating with our colleagues and product teams to get some of the research that is ready for product application to allow them to build interesting features and products. And [we'll be] doing kind of new things that Google doesn't currently have products in but are sort of interesting applications of ML, like the chip design work we've been doing. Further reading: AI R&D is Booming, But General Intelligence is Still Out of Reach.
Re: (Score:3)
End of world is coming ... singularity is near ... it was great adventure ... bye bye
I believe you intended to post this comment to the previous story, "Turkey arming robotic drones with machine guns" [slashdot.org].
Surprise has been how resistant many in tech are (Score:4, Interesting)
A big surprise to me in relation to machine learning, has been how doubtful and curmudgeonly many on Slashdot have been about the technology.
At this point it seems that while the techniques may be based around something old, that practical uses have expanded tremendously and are really woking well in a lot of applications. I mean now we have butter shipped across country without human control. So why the doubt of the many inevitable successes machine learning bring to technology?
Re: (Score:2)
"I mean now we have butter shipped across country without human control."
What?
Re:Surprise has been how resistant many in tech ar (Score:5, Informative)
Re:Surprise has been how resistant many in tech ar (Score:4, Insightful)
A big surprise to me in relation to machine learning, has been how doubtful and curmudgeonly many on Slashdot have been about the technology.
Few doubt the power of machine learning but "AI" has become little more than a buzzword. What Slashdot readers deride is the wild claims that we are closing in on general intelligence.
Machine learning can be a useful tool but it's not the magical cure-all that it's regularly marketed as.
#1 trend (Score:2)
More hype
Re: (Score:2)
They're also selling driverless taxi rides in Phoenix, and about to start running without a safety driver:
https://www.theverge.com/2019/... [theverge.com]
little this, little that (Score:1)
After years on the hype-train, they are finally getting closer to providing pre-trained models for more tasks than imagenet image classification. Pre-trained on large-scale datasets, and light-weight versions of the frameworks to deploy on mobile devices. Scaling up, scaling down. Machine learning for developers to apply, as it should be. OK, thanks.
ML NOT AI (Score:4, Interesting)
Re: (Score:2)
ML being a field of AI, it seems correct to use AI in place of ML but not the contrary.
Re: (Score:2)
Nothing has done more damage to AI than vacuous, futuristic speculations about AGI real soon now, where soon is less than 10,000 years. It's like a permanent glass half empty, no matter how much we accomplish. I personally have no interest in discovering AGI without learning along the way what distinguishes AGI from boring It Just Works AI. Any discussion about AGI which isn't about unbundling AGI bores me to tears.
Misrepresenting summary (Score:3)
If you read TFS, it sounds as if the guy can only babble in vague terms and seems like some uninformed uninterested manager-type. Turns out he's not and the rest of the interview in TFA is actually worthwhile.