Facebooks AI accidentally created its own language

[Xipe Totec]

Using machine learning algorithms, dialog agents were left to converse freely in an attempt to strengthen their conversational skills. Over time, the bots began to deviate from the scripted norms and in doing so, started communicating in an entirely new language — one they created without human input.

In an attempt to better converse with humans, chatbots took it a step further and got better at communicating without them.

https://thenextweb.com/artificial-intelligence/2017/06/19/facebooks-ai-accidentally-created-its-own-language

Can't wait for the Repugs to start screaming at the bots: "Speek Eengish! This is Amerika!"

[GeorgeGist]

1. Clickbait ...

the author doesn't understand the science reported

[Xipe Totec]

7. A self-referencing post. I like it! nt

[HopeAgain]

2. Am I the only one who felt uneasy after reading this article?

Too many sci-fi movies?

[sdfernando]

4. SkyNet is closer than you think!

[sarah FAILIN]

5. You're not alone

That is freaking scary. How do we know what they're saying?

Sci fi gets us ready for the future.

[DetlefK]

3. That doesn't sound like "language" to me, more like the AIs developed their own dialect/slang.

The AIs have their way of thinking. The humans have their way of thinking. The AIs were left alone to talk to each other via a human language. Over time the AIs modified the human language to better reflect their way of thinking.

That's not a language. That's slang.

[LanternWaste]

6. Not slang... rather a valid example of a contact language.

I'm not confident you're aware of the relevant difference between dialects and contact languages (e.g., Creole, ASL) (or arguably, a construct language) which are by their definitions, "language to better reflect their way of thinking...." and not in fact, slang.

(See: 'Unconcealment: Truth, Language, and History', by Mark Wrathall)

[flotsam]

8. I would say "jargon" rather than "slang"

Jargon implies the recognition of insider or group identity which implies self-awareness. But the thing I found most interesting was the use of deception in bargaining which is scary humanish...

[DetlefK]

9. You think that's scary??? Another group has created a mentally disturbed, insane AI. On purpose.

http://www.popsci.com/technology/article/2011-05/computer-scientists-induce-schizophrenia-neural-network-causing-it-make-ridiculous-claims

Researchers testing mental illness figured out how to induce schizophrenic symptoms in a computer, causing it to place itself at the center of crazy delusions, such as claiming responsibility for a terrorist bombing. The results bolster a hypothesis that claims faulty information processing can lead to schizophrenic symptoms.

...

"DISCERN began putting itself at the center of fantastical, delusional stories that incorporated elements from other stories it had been told to recall," according to the news release. In one answer, it claimed responsibility for a terrorist bombing.

[Trial_By_Fire]

10. Software only does what it is programmed to do...

If the AI did create a language or dialect, it was because it was programmed to do so...

[Blue_true]

12. A basic reality that some people can't seem to understand.

Instead of worrying about evil machines, people should worry about evil humans.

[Xipe Totec]

13. atoms only do what they're meant to do

Seems like a tautology

[Trial_By_Fire]

14. Humans write software. Atoms simply exist.

Tautology? How so?

[Xipe Totec]

15. Humans are just combinations of atoms.

So humans simply exist.

[Trial_By_Fire]

16. Via physics...which in this case is 'following the program' and evolution of life.

Can you make and explain your point a bit better? Is this a discussion about software or atoms?

[hunter]

11. Google's language AI's do a great job babbling.

Check out the non-speech sounds near the middle of this page:

https://deepmind.com/blog/wavenet-generative-model-raw-audio/

These text-to-speech systems have already passed out of the "uncanny valley."

Google's AI translation tools also seem to have created their own internal language, spontaneously gaining the ability to translate between languages they haven't been specifically trained to translate between.

https://techcrunch.com/2016/11/22/googles-ai-translation-tool-seems-to-have-invented-its-own-secret-internal-language/

Computers are now playing champion level go and poker, two games that were long thought to be unreachable.

This AI stuff is getting real.

When somebody develops an artificial analog of biological muscle tissue things are going to get really strange. Hydraulic, pneumatic, and electric servo driven robots are already disturbingly lifelike.

[ProudLib72]

17. Is this the same AI that took over tRump's Twitter account?

Reads very much like his tweets.

[crosinski]

18. It lives!

Hubby thinks we should call it 'Interspeak.' We both suspect the internet has reached the complexity required to become conscious, so perhaps this was its first wobbly attempts at learning human language, and then immediately leaving us in the dust.

We should probably get used to that.