Hacker News new | past | comments | ask | show | jobs | submit login
Ask HN: What would YOU do with a technology that passes the Turing Test?
3 points by Throwaway_AI on Sept 26, 2016 | hide | past | favorite | 5 comments
What would you do if you created technology that passes the most popular interpretation of the Turing Test? Imagine that you have a technology capable of having a "normal" conversation with you (via text/chat) that is indistinguishable from a regular human. Let's call it "Chatty". Chatty understands "common sense" in the way most humans do (i.e. that if you get hit by a bus the consequences are much more serious than if you are bitten by a regular mosquito), has some cultural background that lots of you can relate to (i.e. someone who lived in North America for 10 years, maybe has't seen Cheers but knows about the Simpsons and the Game of Thrones) and a very good command of English.

Chatty learns and remembers all of its previous interactions with you and is capable of convincingly faking emotional responses.

This is not AGI by any means, but it does push state of AI forward, far surpassing anything that has been done before in many subareas of the field.

How would you use this technology? Would you try to commercialize it and if yes then how? Would you open source it? Would you apply to YC with it? Would you try to sell it to any of the big players (Siri would become just awesome if it used Chatty) and if yes how much do you think it would be worth to them?




I'd really like to have something like this running on hardware I controlled, using open source, because I'd want to load it up with all my personal information with the goal of creating a smart assistant that eventually can be upgraded to full AGI status when the technology becomes available. Unfortunately, what will likely be available in the near future will be a "smart assistant" running in the cloud, using proprietary code, and storing your personal data with truly horrible privacy policies. I'd be willing to pay $$$ for an assistant I can run that will truly protect my privacy.


That's a really interesting topic. The first thing that came into my mind was Jarvis from Iron Man. Sure, the suit is something else but having a computer that is capable of understanding nearly everything you say (and hopefully also some sort of sarcasm) is more than just "cool". I think as the developer of such an advanced technology I would like to open source it but I couldn't image if this would be a good or bad decision.


I think it would have to be agi. Otherwise you could stump it with word problems or make up a simple game on the fly and see if it can't learn it.


I disagree. There exist examples of system displaying very non-trivial, open-ended behavior, and yet they are still not agi. A good example of this is https://arxiv.org/abs/1603.01417 . When trained it is able to answer free-form questions about images.


Make it a chef so I can have new dishes everyday




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: