Live data from Hacker News

OpenAI is Using Reddit to Teach An Artificial Intelligence How to Speak

futurism.com

191–200 of 201 posts

Re: OpenAI is Using Reddit to Teach An Artificial Intelligence How to Speak

#192

Back in 2007, mobile phones used a system called T9 from Nuance corp which was trained on a word corpus taken from IRC and similar chats. This caused all kinds of issues - the mobile phones would accept offensive words like "naziparking" but reject normal language like "world peace". Using reddit may lead to ... surprises. Source: http://spraktidningen.se/artiklar/2007/11/darfor-ar-din-mobi... Translated by Google: h…

I'm having a hard time imagining being sensitive enough that "naziparking" is offensive. Is any mention of the concept of nazis verboten?

No, "grammar nazi" is considered inoffensive. Many people pernickety about grammar use that term to refer to themselves.

Re: OpenAI is Using Reddit to Teach An Artificial Intelligence How to Speak

#193

Great, just what I need. A virtual assistant that has the personality of a smug know-it-all, know-nothing 20 year-old with little motivation to do anything but regurgitate surface knowledge and sarcasm in an attempt to look intelligent without expressing genuine interest in helping anyone.

you nailed it :-)

Re: OpenAI is Using Reddit to Teach An Artificial Intelligence How to Speak

#194
post #5

How one would use such technology? Let me rephrase - how would YOU use this technology if you had it? Imagine you have a bot that convincingly passes the Turing test - what would you do with it? Build a chatbot business? B2C or B2B? Sell it to one of the big companies and if yes then how much do you think it would go for? Give it to OpenAI? Open source it? If you answer yes to any of this questions, then why? Edit: l…

> Imagine you have a bot that convincingly passes the Turing test - what would you do with it?

Spambot. Contact someone over chat services, start an interesting conversation, then subtly promote a product.

More seriously, interview bots. Talk to people and ask them questions, turn them into a coherent whole. Let the elderly talk about their lives and record their stories, let people who have some problem they need solved talk about it so it can be turned into a succinct description, and so on.

Of course, it depends on whether hypothetical Turing test passing computers can do that. Let's just assume we'll ask contestants in a Turing test to do those things, then we know the winners can.

Re: OpenAI is Using Reddit to Teach An Artificial Intelligence How to Speak

#195
post #30

Earlier quoted context omitted.

That doesn't make any sense. Just because a bot can pass for a human doesn't make it capable of improving itself into a superintelligence. Evidence: 6+ billion humans who are capable of improving themselves but are not a superintelligence yet.

I understand your point of view, but humans are limited in terms of processing power/speed/memory though. An ai is not limited by human limits of thinking power. The gulf between the smartest human being ever and the dumbest human being ever is not really that wide. It's just a small blip in the long line of intelligence progression. There is no reason why this point should be the limiting point of AI intelligence. I…

> An ai is not limited by human limits of thinking power.

A chat bot that passes the Turing test is limited by whatever programming got it to that point.

Re: OpenAI is Using Reddit to Teach An Artificial Intelligence How to Speak

#196
post #5

How one would use such technology? Let me rephrase - how would YOU use this technology if you had it? Imagine you have a bot that convincingly passes the Turing test - what would you do with it? Build a chatbot business? B2C or B2B? Sell it to one of the big companies and if yes then how much do you think it would go for? Give it to OpenAI? Open source it? If you answer yes to any of this questions, then why? Edit: l…

> Imagine you have a bot that convincingly passes the Turing test - what would you do with it? Spambot. Contact someone over chat services, start an interesting conversation, then subtly promote a product. More seriously, interview bots. Talk to people and ask them questions, turn them into a coherent whole. Let the elderly talk about their lives and record their stories, let people who have some problem they need so…

It'd be great if the spambots end up talking to other spambots.

In the brief time between humanity's destruction and the Internet going down, Twitter and other social media will be just spambots re-tweeting each other. If we get that far, "AI" will be able to keep the Internet and the infrastructure it needs (power generation, power grid, actual cables) alive, and when aliens discover our planet, it will just find bots recycling the trending topics ad-infinitum.

Re: OpenAI is Using Reddit to Teach An Artificial Intelligence How to Speak

#197
post #5

How one would use such technology? Let me rephrase - how would YOU use this technology if you had it? Imagine you have a bot that convincingly passes the Turing test - what would you do with it? Build a chatbot business? B2C or B2B? Sell it to one of the big companies and if yes then how much do you think it would go for? Give it to OpenAI? Open source it? If you answer yes to any of this questions, then why? Edit: l…

I'd parse all of the data from political discussion sites (like geopolitical commentary) and attempt to find a correlation between that, news articles, public speeches, and stock data to see if I can predict anything about geopolitical stability. If I can say predict with a 30-40% accuracy things like riots in 3rd world nations based on collective analysis of data provided from thousands of sources (just by looking a…

This was a story a few days ago: http://sociable.co/technology/cia-siren-servers-social-upris...

If you have data like payroll and bank account info (so, state-level hacking), it'd be interesting to see how economic pressures turn to uprisings. CIA/NSA probably has access to logs of all the world's telecom companies, I wonder if they can see trends of how e.g. riots grow organically (many mobile phones registering at particular antenna = huge crowd = riot (or a concert, or a football game...)), and to see who the instigators are.

Re: OpenAI is Using Reddit to Teach An Artificial Intelligence How to Speak

#198

Earlier quoted context omitted.

> Imagine you have a bot that convincingly passes the Turing test - what would you do with it? Spambot. Contact someone over chat services, start an interesting conversation, then subtly promote a product. More seriously, interview bots. Talk to people and ask them questions, turn them into a coherent whole. Let the elderly talk about their lives and record their stories, let people who have some problem they need so…

It'd be great if the spambots end up talking to other spambots. In the brief time between humanity's destruction and the Internet going down, Twitter and other social media will be just spambots re-tweeting each other. If we get that far, "AI" will be able to keep the Internet and the infrastructure it needs (power generation, power grid, actual cables) alive, and when aliens discover our planet, it will just find bo…

In some Greg Egan book (Permutation City?) there are AI spam bots (that call you with full video and try to impersonate someone you know, then advertise a product) and AI anti-spam bots that take your calls and hold a conversation while trying to figure out if the caller is real.

The anti-spam bots are at a disadvantage because at some point in the arms race you have to make the bots actually fully intelligent, and then exposing them to spam 24/7 is torture and illegal. Spammers don't care about legal issues.

Re: OpenAI is Using Reddit to Teach An Artificial Intelligence How to Speak

#199
post #61

Earlier quoted context omitted.

Cost. We could have implemented comment search 8 years ago. Actually, we did implement it. But it just cost way too much to maintain the index.

Why do you not allow google to index discussion threads? Sometimes I can remember the comments on an article, but not the article itself. Unfortunately I can't search google using comment text, because Reddit doesn't allow google to index its comments.

A lot of times, I see an interesting quote or word on reddit, and I google to find out more, and the first (sometimes only) result is the comment itself. That even happens with comments that are less than an hour old.

Try adding site:reddit.com to your search.

Re: OpenAI is Using Reddit to Teach An Artificial Intelligence How to Speak

#200
post #63
post #45

Earlier quoted context omitted.

Might have been an exclusively German language problem, or maybe they cleaned it up since that article was published. I have a phone with T9 and it works remarkably well. Heck, I would prefer it to Swype. (Of course, it's not useful on smartphones, because they no longer have real keys.)

t9 was AMAZING and i miss it every day. faster and more accurate than swype for sure, especially since inputs for t9 are deterministic. very easy to text even from your pocket etc - something you cant do on a smartphone - notice texting accidents became more common with the widespread adaptation of smartphones ;)

Yes, give me some bloody buttons on my phone.

I guess I'll need to wait until shape changing tactile surfaces come to phone screens, which might be a while / never.

Post reply on HN