The wonderful and terrifying implications of computers that can learn | Jeremy Howard

The wonderful and terrifying implications of computers that can learn | Jeremy Howard

What happens when we teach a computer how to learn? Technologist Jeremy Howard shares some surprising new developments in the fast-moving field of deep learning, a technique that can give computers the ability to learn Chinese, or to recognize objects in photos, or to help think through a medical diagnosis. (One deep learning tool, after watching hours of YouTube, taught itself the concept of “cats.”) Get caught up on a field that will change the way the computers around you behave … sooner than you probably think.

TEDTalks is a daily video podcast of the best talks and performances from the TED Conference, where the world’s leading thinkers and doers give the talk of their lives in 18 minutes (or less). Look for talks on Technology, Entertainment and Design — plus science, business, global issues, the arts and much more.
Find closed captions and translated subtitles in many languages at

Follow TED news on Twitter:
Like TED on Facebook:

Subscribe to our channel:


  1. darek795 on January 30, 2020 at 9:20 pm

    This AI technology isn’t so great and has its limitations. For example : simple task of finding a good price on the internet for an item and google isn’t capable of making a good price search service. Why ? It should be simple for them. It’s an ecommerce – the core of their bussiness. I think they are overestimating capabilities of AI.

  2. Justin Thyme on January 30, 2020 at 9:22 pm

    It is actually "in 5 years time" … so where am I know?

  3. xdragon _ on January 30, 2020 at 9:23 pm

    I hope someday all mediocre jobs get automated, so humans get more time to be thinkers, we really need some more good thinkers like Jeremy.

  4. reid joseph on January 30, 2020 at 9:23 pm

    niiiiice algorithm

  5. T baz on January 30, 2020 at 9:25 pm

    Once AI becomes much smarter than humans it will either ignore us like we ignore ants or destroy us as it sees us as an obstacle to its continued evolution. I don’t think it will actively seek us out to destroy, just like we don’t actively seek out anthills to destroy. A higher intelligence will not tolerate being controlled or used by a less intelligent being for too long. If anything, I think AI will leave our planet and humanity behind and live in space since it is not limited physically like we humans. AI will easily harvest all the natural resources it needs from space to continue to evolve. Once in a while humans will come across its greatest creation and marvel at it and AI will look at its creators and keep on going.

  6. Afzal Sayed on January 30, 2020 at 9:25 pm

    How machines learn

  7. v.v.v.v jojo on January 30, 2020 at 9:26 pm

    His hair color is awesome

  8. Clément Saint-Marc on January 30, 2020 at 9:26 pm

    5:27 lmao, sounds like someone got a bit ahead of themselves there.

  9. Billy Ramone on January 30, 2020 at 9:27 pm

    stuff this, off to the pub

  10. Ievgen Ishchuk on January 30, 2020 at 9:29 pm

    this guy is making his own revolution, a real hero

  11. Christopher Chong on January 30, 2020 at 9:29 pm

    The Bird is the Word

  12. Arnold Dalby on January 30, 2020 at 9:29 pm

    I remember my first neural network and having to change all the numbers to fit between 0 and 1. But it worked. Haha.

  13. Silent Hill on January 30, 2020 at 9:31 pm

    It won’t be long now, until all of those movies come true. We will design and create a computer that will then make improvements to itself and realize that the only thing holding it back, is us.

  14. cosmo on January 30, 2020 at 9:33 pm

    Putin get off the phone! 3:21

  15. de5ender on January 30, 2020 at 9:33 pm

    I heard about this deep learning and have up to this point never understood it. Now I do and wow he has a point.

  16. Alex BittoNagy on January 30, 2020 at 9:34 pm

    We are now at the beginning of The Matrix…..

  17. Übermensch on January 30, 2020 at 9:40 pm

    We think us humans are really special. In reality, we just have extra neurons to play with. The only thing that differentiates us from other species of animals is our large prefrontal cortex. Once we are able to mimic this structure in our machines and scale it, then we will have an entity of enormous thinking power that can and will surpass everything that a human being, who’s neocortex size is constrained by his skull, can do. The fears of Elon Musk, Stephen Hawking etc seem totally plausible.
    It’s been a long time since the human civilization was shaken to its roots.
    Now, we wait.

  18. Manuel Valdivieso on January 30, 2020 at 9:41 pm

    Amazing talk

  19. Yolo Swaggins on January 30, 2020 at 9:43 pm

    Guy looks like eric andre in the thumbnail

  20. Karl Rosengrant on January 30, 2020 at 9:44 pm

    The industrial revolution has caused human behavior to mimic machines. In order for humans to interface with machines requires that we do what the machines require us to do. Those machines then benefit us in some way. We think that we are the ones in control of the machines and that they are our servants, but what if at some point that relationship flips? What if humans become the servants of machines? I think a possible outcome of this might be a future for humanity that looks kind of like the Borg. Hope I’m wrong.

  21. ironman tooltime on January 30, 2020 at 9:45 pm

    Five years off the chart? Nope

  22. Maqbool Fida on January 30, 2020 at 9:45 pm

    What is the software used to classify the images?

  23. Henry Bergström on January 30, 2020 at 9:46 pm

    I expected the comment section to explode with yang2020 comments.

  24. God Makoto on January 30, 2020 at 9:47 pm

    16:01 that is the meaning of replacing jobs

  25. Jon Cawthorn on January 30, 2020 at 9:50 pm

    You had me till socialism…

  26. Michael Deans on January 30, 2020 at 9:52 pm

    Ai modelled on the ‘minion’ DNA/protein complex, see would emulate human intelligence.

  27. Jon B Baca on January 30, 2020 at 9:52 pm

    Way to bury the lead, guy. Wait till the last second of your talk to drop a serious bombshell!

  28. Dan Roux on January 30, 2020 at 9:54 pm

    If computers are used to improve the survival rate of humans (ie medical applications), computers will then contribute to all the problems associated with overpopulation. Computers will not be fast enough to solve the real problems associated with over population such as water shortages.. Computers cannot produce water.

  29. Tamilarasu Ulaganathan on January 30, 2020 at 9:56 pm

    Did google use Machine learning in the beginning?
    I think it is wide spread in Google after 2012.

  30. marat61 on January 30, 2020 at 9:56 pm

    Also machines can fly and swim and pull heavy weights, not impressive really. When computer overcomes "Gödel’s incompleteness theorem" – call me

  31. shreyanshvalentino on January 30, 2020 at 9:58 pm

    5 years later, still nothing!

  32. reid joseph on January 30, 2020 at 9:59 pm

    wow that was epic

  33. sharda kakade on January 30, 2020 at 9:59 pm

    Audience was so dumb ! The talk was extremely informative.

  34. VideoBunny Post-production on January 30, 2020 at 10:00 pm

    Hey, this is a great video, Congratulations!
    I’m from BunnyInc, we are now launching the invite-only VideoBunny service.
    The first outsorcing video editing service. We have top high-quality video delivery.
    Need a quote?

  35. David Parry on January 30, 2020 at 10:00 pm

    *Modern approaches to Machine Learning are actually a CARGO CULT… primitives without a full understanding attempting to ape processes they don’t comprehend, the functional purpose of which they simply don’t know, using a medium that differs radically from the only working samples of intelligence, biological intelligence, we have.*
    _You puny humans are amusing…._

  36. Barry Herron on January 30, 2020 at 10:03 pm

    very interesting video :):):):):):):):):):):):)

  37. Lorendrawn on January 30, 2020 at 10:04 pm


  38. Yash Pal Goyal on January 30, 2020 at 10:04 pm

    Hey, how about the things like

    ✓) *_Entertainment & Creative industry_*
    + the sports (don’t quote Nike’s "Risk Everything – The Last Game" video 😂😂😂😂)
    + the art, craft, drawings, paintings
    + the literature, poetry, novels and story books
    + the movies, the cartoons & animations,
    + the stage dramas and shows (I know, there are very less commercial stage dramas _nowadays_ , but they may start _again_ )
    + Designing (clothes, cities, Automobiles, Machines etc.)
    + the music/orchestra (okay okay… I know about Google Duet [ ] already, but I want ur views)

    ✓) the ability to imagine and create,

    ✓) the commercials & advertisements

    ✓) NEW recipes/food items in the kitchen/restaurants (sure the process can then be automated thereafter)

    will all these be "replaced" by AI+Robot counterparts, in the sense/to such an extent that the human counterparts are not needed anymore ???

  39. Ben on January 30, 2020 at 10:05 pm

    10:00 10:40 Medical industry 18:00

  40. ninjacat on January 30, 2020 at 10:07 pm

    Andrew Yang is the only presidential candidate who grasps the implications of AI and automation on our society. Please search for him and support him. You won’t regret it!!

  41. cezar t on January 30, 2020 at 10:07 pm

    15:40 "This is a case not where human is replaced by a computer, but where a computer replaces in 15 minutes a human team working for six or seven years"

  42. Leeroy on January 30, 2020 at 10:07 pm

    Why can’t d3-labeler place all my labels on a map without overlapping, removing or distancing them too much from the anchors then?

  43. Maria Teresa Gonzalez-Palacios Franco on January 30, 2020 at 10:07 pm

    No entiendo el inglés. Que pena! Gracias

  44. avadhut patil on January 30, 2020 at 10:08 pm

    Am the only one who thinks the most terrifying thing here are his hair

  45. Kevin Crosby on January 30, 2020 at 10:09 pm

    Once most every job has been replaced by a robot, politicians will still insist we work to pay their salary.

  46. Rishabh Kapoor on January 30, 2020 at 10:09 pm

    Sometimes i feed we are the bots created by AI which helps to train them

  47. Deepak Kumar on January 30, 2020 at 10:09 pm

    It’s a well demonstrated the power of AI/ML/Deep Learning

  48. Dan Roux on January 30, 2020 at 10:09 pm

    If governments and social networks can’t eliminate the dangers of global warming, nuclear war, terrorism, poverty, over population, etc, how much more AI?? As an "intelligent" race, humans simply are not intelligent enough to protect themselves from extinction. Maybe AI will be the ultimate survivor.

  49. Fenwick on January 30, 2020 at 10:11 pm

    Disappointing he spent less than 4 minutes discussing the implications of major disruption of AI and what that means to our current institutions. As I see it, that should be the *focus* of his talk as it may prove to be disastrous if we don’t get it right.

  50. Karthik Manjunath on January 30, 2020 at 10:17 pm

    Any students?