Do AI chat bots need a personality bypass – or will we only trust gabber 'droids with character?

The speaking phish

> In conclusion, there isn’t any value in bots having unique personalities

I would expect that once a realistic sounding AI starts making cold calls and asking people for their financial secrets, the response rate will go through the roof. The only question would be whether to give it the "personality" of a bank employee, a police officer, a surveyor, one of your friends (with extra input from FB) or an elderly relative in a spot of bother.

But is the technology getting ahead of what users actually want?

Not at all.

The technology is actually falling further and further behind what users actually want.

What users actually want is someone to help them.

Unfortunately that costs money, and that's not what shareholders and CEOs want.

"the AI has to have personality"

It is an effing computer it doesn't have a personality.

When I call a company and get a computer trying to have a conversation with me I find myself being extremely rude, why should I be polite to a machine?

I wonder if the person I end up being put through to gets to hear the conversation first, whatever, expecting an extremely rude Mr Angry will keep them on their toes.

Anybody remember Sirius Cybernetics?

There's been a lot of speculative fiction about bots with personalities. I think all of Asimov's androids essentially had personalities, and of course Douglas Adams posited a misguided corporation that produced far too many things that really shouldn't have had personalities at all.

I can see bots with personalities as being quite interesting as NPCs in various games. There's also the comic, Nonplayer, that's exploring the ramifications of having full AI in an online game world.

A chat bot in a support role doesn't need personality, but it might be nice. Doesn't matter, really, as most of the time you're just reading the manual to somebody. But you could program it with a sexy, alluring voice, though. Or spiff up the jive filter.

by answering more questions

> And IBM has Watson, a machine that famously beat human competitors by answering more

> questions correctly on the American game show Jeopardy.

Hmm, should that be "questioning more answers"? It is "Jeopardy", after all. I guess I'll have to leave it to the AI to decide which is more correct...

"Google’s AI arm DeepMind used WaveNet to make machines sound more natural, and even inserted breathing sounds. "

Great. Expect dirty phonecalls from bored AIs soon.

"To do that, the AI has to have personality. “How do you give a bot personality? It’s simple – you talk to it,” Rodichev said during a presentation at the deep-learning summit."

No, that's how you train a puppy. Unless the entity you talk to actually understands what you tell them, all you do is pile up a heap of data your expert system can work with. Granted, with rising levels of complexity (indexing, references, cross-references, parsing, sorting, collating, what have you) the system will get better and better at mimicking human behaviour. But that system will still have less of a personality than an actual puppy because it still isn't a person.

Okay, creating the equivalent of a highly functional sociopath should be possible.

Logically, this means that the first company functions that an AI could perform far better than any human are not situated at the helpdesk, but in the boardroom.

So, flaws in AI personality are due to human personality flaws? And AI can be taught to act as stupid and pathetic as humans can? Great. AI becomes as screwy as humanity. That's the last thing humanity needs. Instead of focusing on AI and "machine learning", how about focusing on improving human performance, human intelligence, human memory, and human skills? That might actually advance the race of man faster and more accurately than creating Skynet and killing us all just to feed someone's pathetic ego.

Why?

Ok, I get that neural net simulations can be helpful in speech recognition and so on.

But take a hint. Nobody likes talking to machines, not even the trivial, "how may I direct your call" used by banks. I'd far rather use a web browser for banking than talk to a clever voice recognition system. That is a last-resort UI.

Intelligence implies guessing, guessing implies mistakes and for most of the time, we don't want and won't tolerate computers making mistakes, even if they are doing a better job than humans.

What we appear to have is an industry which has run out of ideas for actually making the world a better place, has seen far too many star trek episodes and is obsessed with becoming god! Manufacturing inventions make things better than humans can. IT now seems obsessed with making things a whole lot worse. Worse intelligence than a human, IoT worse than the manual solution, cloud worse than the on-site version, Internet streaming worse than a DVD/Bluray, support moved to increasingly unskilled and unintelligible humans. I've been working for an IT company where the senior engineers have no non-vendor expertise. Great analytical skills, great with the vendor tools, but all at sea when the tools fail. They have no skills in data processing.

Go ahead and and play with your AI, but please don't bring it out of the lab into inappropriate situations.