1) Subscription Division: Ai Tour Guide. Concept wise I think we finally have a winner.
Scale was only ever achievable once the API was in place, which only happened just 3 short weeks ago.
This is the little piece that allows for the launch of the GLOBAL self-service platform and the easy integration into a customer's website.
Without it, it could only have been one customer at a time via our engineers.
The Chairman has said this in the webinars.
In sharing my research, here's the latest article out today.
Friday, 05 July 2024 12:19pmAn employer now using a "digital human" for its onboarding says the move has freed up HR's time while ensuring no crucial information is getting "lost".
Unith, which uses AI to make avatars that organisations can use to represent them, recently started testing its product on its own staff, CEOIdan Schmoraktells HR Daily.
"We took our HR manager Anniek and we cloned her," he says; now all employees are being onboarded by "digital Anniek".
Using its existing product, the process was "extremely simple", Schmorak says. Unith created and uploaded a knowledge base of onboarding information, then used video and audio from the human Anniek to make her digital clone. (Choosing an AI-generated face and voice instead of using a real employee is also an option.)
Now, when new employees meet Unith's HR manager in person, they recognise her from Zoom, having already spoken with a digital version. And while the real Anniek speaks English and Dutch, as a digital human she speaks 60 languages – "digital Anniek is a multi-talent", Schmorak says.
In his view, the trial is proving successful, as before, a lot of onboarding information was getting lost.
"Especially in startups, the information is everywhere," he says, whereas now, "the information is in one place and is being handed over to you, fed with a spoon from a digital version of the onboarder".
"It's actually a great way to get [employees'] attention," Schmorak says, and the HR department receives fewer questions, because "things are much clearer". And if an employee has a question the digital onboarder can't answer, it can still be answered by HR.
"They don't need to stop the process... you can [say] to the digital human, 'Hey, I don't understand this, can I get more information?' and then HR, the real HR, would know how to get back to them."
This is important because a known risk with large language models is that when faced with a question it can't answer, the AI will simply "make stuff up".
But Schmorak says the risk can be mitigated by running the AI in a closed environment. "So we tell the AI, 'Hey, you've got this PDF, whatever in this PDF, that you know, you tell it very accurately. [If] you don't know, you say, 'I do not know'."
He notes that some clients have expressed a preference for an open, ask-me-anything approach from their digital humans – for example, "let them ask about the weather and that's okay" – but he advises against it because "then you employ risk".
Andfor ethical reasons, the people interacting with the AIs are informed that it's a machine, not a human. "We also put other stuff in the experience. So we can put pictures in, we could put graphs, it's not only a human that is in front of you, it's a whole system."
In terms of measuring success, it's early days. But questions being asked include: how many people are logging in; how many people are "actually" using it, and for how long; and where the knowledge base needs to be "beefed up". In some cases, how many questions are coming to a certain division, and whether employee time was actually saved, will also be telling.
The most obvious measure in Unith's trial is that "the real Anniek"no longer spends time onboarding, he says. The entire process is now digitised; she is no longer walking every employee through "the entire thing". There's also a humorous aspect when employees meet, and recognise, the real version.
Schmorak says Unith's clients are yet to use its digital humans for onboarding, but conversations are taking place across multiple industries.
"That's the beauty of a tech platform, of an AI, it can learn anything, assuming you give it the right information."
In areas beyond HR, one of the organisation's biggest clients, a digital health provider, is working towards electronic social workers that speak multiple languages, to interact with some 50,000 patients.
(20min delay)