Did Google Fake Its Duplex AI Demo?
Did Google False Its Duplex AI Demo?
A few weeks ago, at Google I/O, visitor CEO Sundar Pichai demoed an incredible advance in AI and phonation technology. The demo involved a new AI assistant making phone calls to 2 businesses — a hairdresser and a restaurant — to make an appointment and book a reservation, respectively. Yous tin run into the original demo below, but for those of you who don't take access to video, the calls are impressive because the AI responds (and sounds) far more human than your typical voice assistant.
It pauses. It says "Umm," in the aforementioned way a homo might, while because options. When the reservation-taker at the restaurant confuses the AI's request, the AI doesn't first jibbering nonsense. It doesn't even flail. Both demos are impressive, but the restaurant demo is the one, I personally think, that made people sit upwards and blink. But there are also oddities in the recordings that I noticed, even at the time. At present Axios has written a scrap more on the event, including some questions that Google, apparently, doesn't want to answer. Commencement, here's the video:
Note the oddities:
- The businesses never identify themselves
- The humans picking up the phone never give their names
- There's no ambient groundwork noise
- The reservation-takers never asking data. No contact telephone number. No name.
Axios called over two dozen hair salons, including some in Mount View, and every business institution promptly gave its proper noun when contacted. This is simply standard process. It's extremely rare to contact someone at a place of business and not be given either the proper noun of the business, the proper name of the speaker, or both. Even something as unproblematic as "This is Lisa, how may I help you?" tells the person on the other end of the phone that they've reached a visitor as opposed to a personal line. Yet Pichai specifically states: What you lot're going to hear is the Google assistant actually calling a real salon to schedule an date for yous. Let'southward listen.
Google has clammed upwards on the topic. It refuses to say if the calls were edited, even just to remove the business names and identifying information. It refuses to disclose anything about the demo at all.
Does information technology Affair?
At that place are 2 ways to look at this state of affairs. On the one hand, the telephone call applied science was impressive as hell. The AI did answer with believable pauses. It didn't error out when presented with flustered communication. Assuming that the AI was actually generating those responses itself, that'south still an impressive achievement in voice replication (and nosotros've covered Google'due south piece of work in this area earlier).
But Sundar Pichai didn't pitch this demo every bit a genuinely impressive advance in vocalization applied science. He was demoing (or claiming to demo) a situation in which an AI assistant could respond to unexpected chat prompts, confusion, and real-world scenarios. And it's entirely possible that the reason Google partially or completely staged the demo (and therefore lied in its framing of the presentation) is because information technology can't guarantee that the person on the other end of the phone will take an emphasis that its AI tin can understand. Information technology may not be able to guarantee that the conversation won't take a plow that its AI can handle. And it may not be able to promise that its AI will understand the information provided depending on how that information is verbally ordered. This blazon of "fuzzy" processing is something human brains are very, very good at, and it's an expanse where AI has generally struggled.
But hither'south one affair nosotros do know. When companies pull off major breakthroughs, especially in an surface area like artificial banana engineering, they typically can't wait to show them off in as many scenarios equally possible. When Apple launched Siri or Microsoft launched Cortana, they showed off their respective capabilities at great length. To this 24-hour interval, Microsoft pushes regular Cortana news about the features and capabilities of the platform.
All we take on Google Duplex, in contrast, is ane canned AI demo shown at a Google event, under conditions that raise existent questions nigh whether or non the result was staged. Google'due south refusal to answer questions about its demo doesn't leave it looking good in this scenario. If Google actually staged this event, then we withdraw our earlier remarks. The visitor didn't laissez passer the Turing Test at all. It just demonstrated a theoretical scenario that might one twenty-four hour period atomic number 82 to such an accomplishment.
Source: https://www.extremetech.com/computing/269497-did-google-fake-its-google-duplex-ai-demo
Posted by: greenguaraction.blogspot.com

0 Response to "Did Google Fake Its Duplex AI Demo?"
Post a Comment