The Real Reason to be Afraid of Artificial Intelligence | Peter Haas | TEDxDirigo

[angelorivera]

He's warning the people, by educating them. The audience thinks it's entertainment but that's what he's doing( educating them)
to open their eyes!

[markbrantingham]

Once you've seen enough of human nature, you'll know that there is absolutely no chance that humans are not going to develope AI until it gets out of control.

[jeremyanderson]

The most true thing he said in the whole video - "you may not care about this because you arent facing criminal sentencing". That sentence sums up most human sentiment, and its sad.

[markmarsh]

That was BRILLIANT ... and 4 years later, which is centuries in terms of AI evolution, he's still EXACTLY right.

[vanhalenps]

Video works fine on 1.5x speed.. Use your extra 4 minutes wisely

[jgrab]

"The AI algorithm conflated the absence or presence of snow with the absence or presence of a wolf." —That explains a lot of my Amazon "For You" recommendations.

[Ratplague]

Don't fear the robots, fear the people who OWN the robots. Fear the people who OWN the data.

[jpheals]

Great talk my friend ! "What we must fear is our own laziness !" That says it all.

[qrprisq]

A man of reason, I'm glad somebody in the robotics field (GETS IT) !!!

[cupofkoa]

This is the exact type of thinking we need to go forward. Very sensible and thoughtful.

[monkeymind]

Very good. People should listen to this now. It is getting crazier out there, and we did not pause to think of the consequences.

[suesheification]

This is exactly why I fear people more than AI. Most people might as well be AI with their lack of ability to critically reason.

[kurtjensen]

Best words spoken in Jurassic Park by Jeffrey “ you kept wondering if you could but did not ask your self if you should!”

[ahyaok]

"Even the developers who work on this stuff have no idea what it's doing." So true.

[meteorable]

In the 70s when I was college student a statistics professor lectured about "error". This lecture changed my life for the better in ways far removed from statistics.
Of course, the definition of error commonly depends on subjective matters....And, that is what worries me, but not necessarily you.
AI worries "me".... but what do I know?

[aybee]

3:55 and there's the rub! "Even the developers who work on this stuff, have no idea what it's doing!"

[kaiyote]

Great talk. Finally someone able to break down the specific issues with these systems

[grupoelcomienzo]

Man! You nailed it. Please continue your AI awareness speaches

[garyclouse]

A few years ago, I worked with a commercial ai package that used a neural network simulation to implement fuzzy logic in handwriting recognition.. The first thing I noticed was that, by giving the computer the ability to make educated guesses, the computers also gained the ability to make mistakes. So the AI algorithm could be tuned by specifying a confidence level. If the AI's confidence was below that level, it would request human intervention.

[rickycollins]

Hands down one of the most terrifying talks i have ever heard. So many facets to take into account with AI, all human.