AI can now have a seamless audio conversation, and enterprise AI is being tailor-made to take over HR tasks. HRM looks at the unveiling of several projects that could impact work.

It may end up being the “Mm-hmm” heard around the world. In a demonstration at Google I/O, the CEO of the tech giant showed how the company’s signature assistant has learned how to phone a hair salon and book an appointment. Notably, the person taking the appointment never suspects they’re talking to a machine.

In fact the voice is so realistic, and so capable of dealing with the difficult little intricacies of our conversations, that it responds to “Sure, give me one second” with a very human murmur of assent. The audience at the presentation couldn’t help but gasp and laugh (check it out, and be equal parts impressed and shaken).

As with any tech announcement, it’s important to be realistic. As explained on Google’s AI blog the assistant is limited to hair appointments, restaurant bookings, and checking the hours of operation during a holiday. Plus the assistant (which is only just being beta-tested) will launch with a self-monitoring capability that will mean it brings in a human operator when it can’t complete the task.

On the other hand, you also shouldn’t be too sceptical. At the moment the AI is quite capable of handling someone who changes their mind mid-sentence, knows how to elaborate on a point, and can handle interruptions. These are big milestones in having AI conversations with humans, and the blog link above has the audio to prove it. It’s a step toward a future where our interaction with computers becomes voice-first.

The future voice of HR

For HR, it’s not hard to see how Duplex could take over aligning their work appointments and, further down the line, more intricate conversations.

For instance, you can see how this AI could conduct a reference check. While such a call is more complicated than booking an appointment (and wouldn’t be happening any time soon) it does have the kind of defined outcomes that make it easier for AI to learn.

A reference check looks to ascertain specific information such as capabilities and qualifications. Conversations about cultural fit would be more difficult – as there’s no evidence the AI can pick up on tone and confidence levels – but less complicated checks could conceivably be in its wheelhouse.

Indeed, HR professionals might be brought on board to train the AI in such conversations. As explained on the blog “To train the system in a new domain, we use real-time supervised training. This is comparable to the training practices of many disciplines, where an instructor supervises a student as they are doing their job, providing guidance as needed.”

AI for your meetings, and for employee training

Not to be outdone by its rival, Microsoft has recently unveiled AI applications that are tailored for work.

The company’s work collaboration software, Teams, is getting upgrades this year that will allow it to seamlessly record and transcribe meetings for later search-enabled reference. But more than that, say you realise in a meeting that you need a manager to be involved with what’s going on. You can, with your natural voice, tell the AI assistant and it will call them for you.

On a more HR specific front, as part of its push towards focusing on businesses, Microsoft introduced Marcel, a large AI, cloud computing and enterprise software project. It’s in partnership with the world’s third biggest advertising company, Publicis. The aim is to get the 80,000 employees of the French multinational to be more collaborative and productive by breaking down the silos in an organisation that’s really more of a holding company for lots of advertising companies.

The three apps revealed in the initial announcement of Marcel are, according to TechCrunch:

Daily Six: a training/engagement app sources six pieces of content every day that are specific to the employee using the app. According to TechCrunch the content includes “updates on clients, suggestions for creative activities, and reading recommendations”. The criteria is based on both what the staff member is currently working on and what they will be working on.

Open Brief: an app will enable clients to request work or recommendations from the whole of Publicis – so not just a single organisation but to every organisation within the parent company. It will use AI to target those specialists who will be most helpful.

Expert Match: an app that will boost in-company mentoring. Like Open Brief but for employees, staff can use it to connect easily with anybody who is able and willing to help out, whether that be with a single question or a request for a mentoring relationship. Again, AI will pick up on clues within requests, and better target those who will hear requests.

Intriguingly, none of these apps  help write ads. They all take on what could be considered HR tasks. When dealing with 80,000 employees you can see why AI would be more effective at making connections than an HR team – it’s cheaper, and what it lacks in nuance it makes up for in its ability to track volume. “It will help us to leverage the talent in the room, tapping people who deserve to do and grow more,” CEO and chairman of Publicis, Arthur Sadoun told TechCrunch.