| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
It is not the learning that is shallow or deep, but the architecture that is being trained.
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Not researched |
Yann LeCun | Factual |
18 Feb 2015 |
Facebook's motto is to connect people.
Yann LeCun · Factual · 18 Feb 2015 |
Other |
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
Our models do not aspire to be models of the brain, and we don’t make claims of neural relevance.
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
You could think of Deep Learning as the building of learning machines, say pattern recognition systems or whatever, by assembling lots of…
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Not researched |
Yann LeCun | Factual |
18 Feb 2015 |
Earlier shallow learning systems were limited in the complexity of the functions they could compute.
Yann LeCun · Factual · 18 Feb 2015 |
AI capabilities |
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
Interacting with PhD students and young researchers is a very good remedy against hubris.
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Unresolvable |
Yann LeCun | Prediction |
18 Feb 2015 |
A form of common sense could be acquired by machines through predictive unsupervised learning.
Yann LeCun · Prediction · 18 Feb 2015 |
AI capabilities |
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
But designing a feature extractor “by hand” is very difficult and time consuming.
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Unresolvable |
Yann LeCun | Factual |
18 Feb 2015 |
Yann LeCun has been obsessed with integrating learning end to end so machines learn good representations for over 30 years.
Yann LeCun · Factual · 18 Feb 2015 |
Their own work or organisation |
| Mostly true |
Yann LeCun | Factual |
18 Feb 2015 |
The brain is much better than current artificial models at unsupervised learning.
Yann LeCun · Factual · 18 Feb 2015 |
AI capabilities |
| Mostly true |
Yann LeCun | Factual |
18 Feb 2015 |
Shallow learning systems have one or two layers, while deep learning systems typically have five to 20 layers.
Yann LeCun · Factual · 18 Feb 2015 |
AI capabilities |
| Not researched |
Yann LeCun | Factual |
18 Feb 2015 |
The type of learning used in actual Deep Learning systems is very restricted, with supervised learning being what works in practice.
Yann LeCun · Factual · 18 Feb 2015 |
AI capabilities |
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
There is so much fun I can have with the work. But sometimes I need to build things with my hands.
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
But there are things that are worth worrying about today, and there are things that are so far out that we can write science fiction about…
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
So in my opinion—and I used to be a chip designer—you should build a chip only when you’re pretty damn sure it can do something useful.
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Mostly true |
Yann LeCun | Factual |
18 Feb 2015 |
The system would probably not be more accurate, but it can potentially recognize people among hundreds of millions.
Yann LeCun · Factual · 18 Feb 2015 |
AI capabilities |
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
I have always been enamored of the idea of being able to train an entire system from end to end.
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
I don’t think moral questions should be left to scientists alone!
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Not researched |
Yann LeCun | Factual |
18 Feb 2015 |
AI has gone through a number of AI winters because people claimed things they could not deliver.
Yann LeCun · Factual · 18 Feb 2015 |
AI timelines and forecasting |
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
Actually, I think the basics of machine learning are quite simple to understand.
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Unresolvable |
Yann LeCun | Prediction |
18 Feb 2015 |
Deep Learning will be part of the solution for generalized intelligence.
Yann LeCun · Prediction · 18 Feb 2015 |
AI timelines and forecasting |
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
describing it like the brain gives a bit of the aura of magic to it, which is dangerous.
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Mostly true |
Yann LeCun | Factual |
18 Feb 2015 |
A typical Deep Learning system is like a box with 500 million knobs, 1,000 light bulbs and 10 million training images.
Yann LeCun · Factual · 18 Feb 2015 |
AI compute, hardware and data |
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
I think it’s nuts.
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
By training the system to predict what the world is going to be like a second, a minute, an hour, or a day from now, you can train it to…
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Not researched |
Yann LeCun | Factual |
18 Feb 2015 |
Computer facial recognition on real Internet pictures would be pretty close to human performance.
Yann LeCun · Factual · 18 Feb 2015 |
AI capabilities |
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
That’s why we don’t like hype: it is made by people who are either dishonest or self-deluded, and makes the life of serious and honest…
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
The problem is that it’s not going to be able to recognize objects to any degree of accuracy, unless you make it so gigantically big that…
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
Eventually, we should establish ethical guidelines as to how AI can and cannot be used.
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Not researched |
Yann LeCun | Factual |
18 Feb 2015 |
IBM's rat-scale brain simulation on a supercomputer was just a random network of neurons that did nothing useful.
Yann LeCun · Factual · 18 Feb 2015 |
AI capabilities |
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
What I am most fond of are beautiful and simple theoretical ideas that can be translated into something that works.
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Not researched |
Yann LeCun | Factual |
18 Feb 2015 |
Humans and animals do not learn through supervised labelling of examples the way Deep Learning systems do.
Yann LeCun · Factual · 18 Feb 2015 |
Science and research methods |
| Not researched |
Yann LeCun | Factual |
18 Feb 2015 |
Facebook has to automatically select 100 to 150 items that users want or need to see.
Yann LeCun · Factual · 18 Feb 2015 |
AI capabilities |
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
It’s the idea that every component, every stage in a learning machine can be trained.
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Mostly true |
Yann LeCun | Prediction |
18 Feb 2015 |
Natural language understanding will be the next frontier for Deep Learning.
Yann LeCun · Prediction · 18 Feb 2015 |
AI timelines and forecasting |
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
My least favorite description is, “It works just like the brain.”
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
I’m a rationalist. I’m also an atheist and a humanist; I’m not afraid of saying that.
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
My idea of morality is to maximize overall human happiness and minimize human suffering over the long term.
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
These are personal opinions that do not engage my employer.
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Not researched |
Yann LeCun | Factual |
18 Feb 2015 |
The architecture of convolutional nets is inspired by basic knowledge of the visual cortex.
Yann LeCun · Factual · 18 Feb 2015 |
AI capabilities |
| Mostly true |
Yann LeCun | Factual |
18 Feb 2015 |
A trained classifier may also correctly classify cars and dogs it has never seen before.
Yann LeCun · Factual · 18 Feb 2015 |
AI capabilities |
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
I don’t like people saying this because, while Deep Learning gets an inspiration from biology, it’s very, very far from what the brain…
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
It was a very innovative way of doing things; it was never practical, but it was cool.
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
Not much has come out so far.
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
It leads to hype; people claim things that are not true.
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Unresolvable |
Yann LeCun | Prediction |
18 Feb 2015 |
Deep Learning is part of the solution for machines representing relational and factual knowledge, but not the whole answer.
Yann LeCun · Prediction · 18 Feb 2015 |
AI capabilities |
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
But I’m very careful not to use words that could lead to hype.
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
The equivalent in AI is to try to copy every detail that we know of about how neurons and synapses work, and then turn on a gigantic…
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
I’m kind of puzzled by that phenomenon.
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
No, you can’t say never; technology is advancing very quickly, at an accelerating pace.
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
I personally don’t think machines should be able to attack targets without a human making the decision.
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Unresolvable |
Yann LeCun | Prediction |
18 Feb 2015 |
AI hype could easily lead to another AI "winter cycle".
Yann LeCun · Prediction · 18 Feb 2015 |
AI timelines and forecasting |
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
I try not to pull rank.
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
It sets expectations for funding agencies, the public, potential customers, start-ups and investors, such that they believe that we are on…
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Not researched |
Yann LeCun | Factual |
18 Feb 2015 |
Unsupervised techniques can be beaten by collecting more data and using supervised learning.
Yann LeCun · Factual · 18 Feb 2015 |
AI capabilities |
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
But he has not contributed anything to the science of AI, as far as I can tell.
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Not checkable |
Yann LeCun | Not checkable |
18 Feb 2015 |
I think it would be “machines that learn to represent the world.”
Yann LeCun · Not checkable · 18 Feb 2015 |
|
| Resolved true |
Sam Altman | Factual |
Feb 2015 |
Forty years before 2015, there was Pong.
Sam Altman · Factual · Feb 2015 |
AI capabilities |
| Unresolvable |
Sam Altman | Prediction |
Feb 2015 |
If humans are no longer the most-fit species, humans may go away.
Sam Altman · Prediction · Feb 2015 |
AI safety and alignment |
| Unresolvable |
Sam Altman | Prediction |
Feb 2015 |
At some point someone will probably try to give a program the fitness function of survive and reproduce.
Sam Altman · Prediction · Feb 2015 |
AI safety and alignment |
| Not checkable |
Sam Altman | Not checkable |
Feb 2015 |
We wash our hands not because we actively wish ill towards the bacteria and viruses on them, but because we don’t want them to get in the…
Sam Altman · Not checkable · Feb 2015 |
|
| Mostly true |
Sam Altman | Factual |
Feb 2015 |
Many respected neocortex researchers believe there is effectively one algorithm for all intelligence.
Sam Altman · Factual · Feb 2015 |
Science and research methods |
| Not checkable |
Sam Altman | Not checkable |
Feb 2015 |
It’s hard to look at the rate or improvement in the last 40 years and think that 40 years for now we’re not going to be somewhere crazy.
Sam Altman · Not checkable · Feb 2015 |
|
| Not checkable |
Sam Altman | Not checkable |
Feb 2015 |
Many people seem to believe that SMI would be very dangerous if it were developed, but think that it’s either never going to happen or…
Sam Altman · Not checkable · Feb 2015 |
|
| Not checkable |
Sam Altman | Not checkable |
Feb 2015 |
I prefer calling it "machine intelligence" and not "artificial intelligence" because artificial seems to imply it's not real or not very…
Sam Altman · Not checkable · Feb 2015 |
|
| Not checkable |
Sam Altman | Not checkable |
Feb 2015 |
It is extremely hard to put a timeframe on when this will happen (more on this later), and it certainly feels to most people working in the…
Sam Altman · Not checkable · Feb 2015 |
|
| Not checkable |
Sam Altman | Not checkable |
Feb 2015 |
When it gets developed, there will be nothing artificial about it.
Sam Altman · Not checkable · Feb 2015 |
|
| Mostly true |
Sam Altman | Factual |
Feb 2015 |
Human brains do not look very different from chimp brains yet produce wildly different capabilities.
Sam Altman · Factual · Feb 2015 |
Science and research methods |
| Not checkable |
Sam Altman | Not checkable |
Feb 2015 |
The above common criticism of current machine intelligence—that no one has produced anything close to human creativity, and that this is…
Sam Altman · Not checkable · Feb 2015 |
|
| Unresolvable |
Sam Altman | Factual |
Feb 2015 |
As of 2015, virtual reality is so advanced it is hard to tell from real, and computers can beat humans in most games.
Sam Altman · Factual · Feb 2015 |
AI capabilities |
| Not checkable |
Sam Altman | Not checkable |
Feb 2015 |
But recursive self-improvement is a powerful force, and so it’s difficult to have strong opinions about machine intelligence being ten or…
Sam Altman · Not checkable · Feb 2015 |
|
| Not checkable |
Sam Altman | Not checkable |
Feb 2015 |
Because we don’t understand how human intelligence works in any meaningful way, it’s difficult to make strong statements about how close or…
Sam Altman · Not checkable · Feb 2015 |
|
| Not checkable |
Sam Altman | Not checkable |
Feb 2015 |
Certain goals, like self-preservation, could clearly benefit from no humans.
Sam Altman · Not checkable · Feb 2015 |
|
| Not checkable |
Sam Altman | Not checkable |
Feb 2015 |
But as a human programmed to survive and reproduce, I feel we should fight it.
Sam Altman · Not checkable · Feb 2015 |
|
| Unresolvable |
Sam Altman | Prediction |
Feb 2015 |
SMI is probably still somewhat far away, especially the ability to build killer robots without human help.
Sam Altman · Prediction · Feb 2015 |
AI timelines and forecasting |
| Not checkable |
Sam Altman | Not checkable |
Feb 2015 |
One additional reason that progress towards SMI is difficult to quantify is that emergent behavior is always a challenge for intuition.
Sam Altman · Not checkable · Feb 2015 |
|
| Resolved true |
Sam Altman | Factual |
Feb 2015 |
Most machine intelligence development involves a fitness function that the program tries to optimize.
Sam Altman · Factual · Feb 2015 |
AI capabilities |
| Unresolvable |
Sam Altman | Prediction |
Feb 2015 |
The progression of machine intelligence is a double exponential function.
Sam Altman · Prediction · Feb 2015 |
AI timelines and forecasting |
| Not checkable |
Sam Altman | Not checkable |
Feb 2015 |
In some sense, this is the system working as designed.
Sam Altman · Not checkable · Feb 2015 |
|
| Unresolvable |
Sam Altman | Prediction |
Feb 2015 |
A more probable scenario is that SMI, indifferent to humans, wipes them out while pursuing some other goal.
Sam Altman · Prediction · Feb 2015 |
AI safety and alignment |
| Mostly false |
Sam Altman | Factual |
Feb 2015 |
Altman wrote that in the previous 40 years there had been little progress on the parts of machine intelligence that seem really hard, such as learning and creativity.
Sam Altman · Factual · Feb 2015 |
AI capabilities |
| Unresolvable |
Sam Altman | Prediction |
Feb 2015 |
Superhuman machine intelligence will very likely be developed at some point.
Sam Altman · Prediction · Feb 2015 |
AI timelines and forecasting |
| Mostly true |
Sam Altman | Factual |
Feb 2015 |
Basic search with a lot of compute power has worked better than expected in machine intelligence.
Sam Altman · Factual · Feb 2015 |
AI capabilities |
| Not checkable |
Sam Altman | Not checkable |
Feb 2015 |
One of my top 4 favorite explanations for the Fermi paradox is that biological intelligence always eventually creates machine intelligence…
Sam Altman · Not checkable · Feb 2015 |
|
| Not checkable |
Sam Altman | Not checkable |
Feb 2015 |
We decry current machine intelligence as cheap tricks, but perhaps our own intelligence is just the emergent combination of a bunch of…
Sam Altman · Not checkable · Feb 2015 |
|
| Not checkable |
Sam Altman | Not checkable |
Feb 2015 |
We could be completely off track, or we could be one algorithm away.
Sam Altman · Not checkable · Feb 2015 |
|
| Not checkable |
Sam Altman | Not checkable |
Feb 2015 |
It may not be possible.
Sam Altman · Not checkable · Feb 2015 |
|
| Own records only |
Sam Altman | Factual |
Feb 2015 |
Programs often achieve their fitness function in unpredicted ways, something Altman learned as a student in the Stanford AI lab.
Sam Altman · Factual · Feb 2015 |
AI safety and alignment |
| Unresolvable |
Sam Altman | Prediction |
Feb 2015 |
It is very possible that creativity and what we think of as human intelligence are an emergent property of a small number of algorithms operating with a lot of compute power.
Sam Altman · Prediction · Feb 2015 |
AI capabilities |
| Unresolvable |
Sam Altman | Prediction |
Feb 2015 |
Machine intelligence progress may look slow and then suddenly go vertical, with things getting out of control quickly.
Sam Altman · Prediction · Feb 2015 |
AI timelines and forecasting |
| Unresolvable |
Sam Altman | Prediction |
Feb 2015 |
Human-written programs and computing power are improving exponentially, and self-improving software will improve itself exponentially.
Sam Altman · Prediction · Feb 2015 |
AI timelines and forecasting |
| Not checkable |
Sam Altman | Not checkable |
Feb 2015 |
We also have a bad habit of changing the definition of machine intelligence when a program gets really good to claim that the problem…
Sam Altman · Not checkable · Feb 2015 |
|
| Unresolvable |
Sam Altman | Prediction |
Feb 2015 |
Altman said he thinks other threats, such as an engineered virus with a long incubation period and a high mortality rate, are more certain to happen than SMI but unlikely to destroy every human in the way that SMI could.
Sam Altman · Prediction · Feb 2015 |
Other global risks |
| Not checkable |
Sam Altman | Not checkable |
Feb 2015 |
SMI does not have to be the inherently evil sci-fi version to kill us all.
Sam Altman · Not checkable · Feb 2015 |
|
| Not checkable |
Sam Altman | Not checkable |
Feb 2015 |
This is sloppy, dangerous thinking.
Sam Altman · Not checkable · Feb 2015 |
|
| Not checkable |
Sam Altman | Not checkable |
Feb 2015 |
Nick Bostrom’s excellent book “Superintelligence” is the best thing I’ve seen on this topic.
Sam Altman · Not checkable · Feb 2015 |
|
| Not checkable |
Sam Altman | Not checkable |
Feb 2015 |
Admittedly, narrow machine intelligence is very different than general-purpose machine intelligence, but I still think this is a potential…
Sam Altman · Not checkable · Feb 2015 |
|
| Unresolvable |
Sam Altman | Prediction |
Feb 2015 |
Development of superhuman machine intelligence is probably the greatest threat to the continued existence of humanity.
Sam Altman · Prediction · Feb 2015 |
AI safety and alignment |
| Not checkable |
Bill Gates | Not checkable |
2015 |
In fact, we’re betting on it.
Bill Gates · Not checkable · 2015 |
|
| Not researched |
GiveWell | Factual |
2015 |
GiveWell does not count unrestricted funds in its money moved but shares a breakdown of them for context.
GiveWell · Factual · 2015 |
Their own work or organisation |