Facebook scales back AI flagship after chatbots hit 70% f-AI-lure rate So it begins. Facebook has scaled back its ambitions and refocused its application of "artificial intelligence" after its AI bots hit a 70 per cent failure rate. Facebook unveiled a bot API for its Messenger IM service at its developer conference last April. Facebook CEO Mark Zuckerberg had high hopes. TenCent's WeChat was …
So the messages got interrupted by "relevant links" and were a failure as to relevance and just about anything else. The problem for FB is that this hits them in the revenue stream... ads that everyone ignores. The second problem is that it tells users... err.... "monetized resources" who might wake up and realize that those quiet chats and private messages aren't. Most may realize that FB slurps everything but this points it out to them in a way they probably won't like much.
> "when Facebook tries to automate its systems, things always go wrong. The problem is not that Facebook does not have the right people but..."
You know, anywhere else, if the thing I was hired to work on kept bursting into flames people would quickly deduce that I'm not the right person for the job. And yet here we are with a Facebook apologist too enamored with the too-big-to-fail website that he can't bring himself to admit that the problem really is the people. The people who trusted algorithms to solve social problems. The people who turned a blind eye to its treatment of minorities and the vulnerable. Their push to monetize by forcing people to use "real" names, with an arbitrary definition of what "real" is -- inevitably leading to disaster when it bumped into another poorly-implimented technology: Their abuse system. And then the rise of fake news, after they got rid of the humans curating it and replaced it with algorithms that quickly went awry and blew up a major election.
And never, ever, not once, has Facebook or its apologists ever taken responsibility for this. And their reaction is downright childish: "The algorithm did it." This is like coming home to find the livingroom lamp smashed and all the kids saying "Somebody else did it." Right. It just fell over on its own. It wasn't at all that you were tossing a football around and it smacked into it. I mean, it wasn't your hands that pushed it over so it wasn't you, right? No, it was the ball. It had it in for that lamp.
The "social problem" is that Facebook exists.
The solution to a high percentage of Internet scams, bullying, exploitation of personal information, fake news etc is simply to turn off Facebook. It provides nothing for the public that isn't available more privately (or publicly), for free, just as easy to use, elsewhere.
And then the rise of fake news, after they got rid of the humans curating it and replaced it with algorithms that quickly went awry and blew up a major election.
Unless you are alluding to some facebook moderator election I think you are just repeating bad talking points.
is that it is largely irrelevant. We've had the ability to craft text and email groups for decades and you know what? NO ONE DOES IT. Why? Because it's overkill and painfully distracting and ultimately unnecessary. Unless you are noteworthy enough to have your own newsletter, the most practical use of a group message is to invite people to an event. Which there are dedicate services for, so what is Facebook for again? Postings shitty pictures of your ugly kids for everyone to Like, or their not your friends? News? If Facebook were to close down today the only people to complain would be Facebook employees who don't understand that their company produces no real product and very dubious services. Oh, and criminals who use it to find marks. There is no practical use for a Facebook account, and ultimately a Facebook service. Idiots made Zuck a household name and a billionaire. I still don't see why.
>it's overkill and painfully distracting and ultimately unnecessary.
I disagree - and I'm someone who can't stand fb and doesn't have an account.
What facebook does provide is an authenticated presence. Not authenticated in terms of who you are, but authenticated in terms of, you're the same person attached to that facebook account as you were yesterday. - at least for people who already know each other offline.
It is a bit like a telephone number - once you know the right one, you can call it and be reasonably sure of the person picking up.
People use fb like an address-book + group chat. It's power is in the data that is already in there - the pictures that we assume will be around forever, but won't be.
What we are in desperate need of is a point of presence protocol and sophisticated address-book and identity management. We need to ditch single-identity communications systems, such as skype.
I think we'll need to see a revenue collapse at facebook which leads to service degradation, before anything changes. There's too much historical data that people want to keep in fb for them to move to another service just because its trendy.
Well said and I would go half a league further - FB is wasteful and counter-productive. It has reached epidemic proportions now, as I look at a website - news, manufacturer, what-have-you - and there is a box telling me to visit their FB page. Why TF would I want to go somewhere else when I am already looking at the main website? Mad. Just imagine the resources including internet hardware to handle such nonsense.
Some experts fretted that if China opened up a lead in AI, the West would be doomed to lose World War 3. Others suggested that whichever superpower lost the AI arms race would relapse into a state of primitive technology feudalism.
Woah the hype of the Cold-War "5th Generation" is back. I remember well. "We are doomed! QUICK, REAGAN, GIMME MONEY!"
Back then the press was generally against WW3 though. Nowadays, it seems so-called "journalists" ejaculate heavily every single morning over the thought of Tank War Europe ("the nuclear deal") or "Limited Nuclear Exchange And Then We Win" in the Pacific.
I'm eagerly awaiting further developments. It's gonna be hella rad.
What we see in our current project (www.chatler.ai) is that AI can reach easily certain level of confidence in answers, and can reduce number of possible answers. But to reach 100% confidence in complex conversations is almost impossible. Thats why we're combining chatbot like functions with human supervision. AI just recommends options and humans select the best one. This way we can help chat agents to chat more productively.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
