Key Idea 9 - your agentic AI will need to authenticate you when you talk to them
The title's a mouthful, but it's made up of several key points; let's unpack it.
Last week my colleague Cliff Mann illuminated how, despite all of our assumptions, the voice remains a key mode of interaction as people still call businesses over the phone.
One of Cliff's strongest arguments was that with advances in new communications media, the role of voice has transformed, as opposed to being rendered obsolete.
But it still goes without saying that in our daily lives, we largely interact with the online world using our fingertips, whether it's via physical keyboard & mouse (vintage!) or via touchscreen & virtual keyboard.
As the world increasingly shifts to AI agents entering our daily lives, their integration into our habits and routines must become increasingly smooth, seamless. One of the ways this will happen, according to Microsoft's Satya Nadella , is by having the AI agents talk to us, and us in turn talk to them.
From a practical standpoint, it also begins to make a lot of sense that voice will feature prominently in our interactions with agentic AI. As the tasks and responsibilities assigned to AI become more complex, we will benefit from natural, real-time forms of expression, foremost of which is the voice.
In the future, people will speak (to their AI personal assistants)
Voice is also portable, and doesn't require you to be positioned at or very near specialized hardware. You can walk from room to room in your house, carrying on a conversation with your AI agent while doing chores.
But even for tasks you're performing on the computer, the AI agent is the friend you're showing something to. You want to point at an Excel formula and ask it for help.
You want it to help you refine a web search after you give it your initial comments on the results you're looking at.
You'll also be working on the video you want to post to TikTok or YouTube (another place we'll be using our voices in the future!), and maybe you need to tell your agent where you want the cuts to be in the shots you're compositing, or what kind of adjustments to make to video effects or music.
The best user experience here is probably to be able to stay focused on the primary interface (image, video, Excel sheet, etc) while giving the agent your comments and instructions. In this setting the best way to convey your thoughts is by using your own voice, in your own words.
领英推荐
You'll just talk to the agent while performing tasks, and it will help you figure things out in real-time.
If things do actually develop as predicted by the tech giants, the AI agents will also be acting on your behalf, to perform complex and sensitive tasks.
Making purchases for you? Paying your bills? Making trades in your investment portfolio? Coordinating schedules with family members? Scheduling doctor's appointments?
There are a multitude of things we spend time doing that we might like for a personal assistant to do, if we could afford one. But these are all incredibly sensitive tasks.
Your AI personal assistant will be able to act on your behalf, even for sensitive tasks such as banking or booking appointments
The agentic AI, the personal assistant for the rest of us, will need to trust that it's receiving instructions from you, yourself, if it's to do anything on your behalf.
And since you're already talking to it, we can use your voice to authenticate you.
Of course the fraudsters will employ various tactics, including the use of deepfakes, to try and fool the agentic AI systems. But we've got that covered too.
Your AI personal assistant (and AI agents in general) needs to be sure it is talking to the real YOU, and must therefore be able to authenticate you
Customer service channels (including by voice) will also be transformed by agentic AI, as discussed in this Substack post. Interestingly, the authors indicate that the only remaining gap for AI in customer service is in being able to identify and authenticate users.
The safe way to deploy agentic AI is by loading it up with strong and robust authentication technology. And as long as we're talking to our AI personal assistants of the future, they'll need to be able to use voice to make sure they're speaking with the real you.
Since your AI personal assistant will converse with you, it might as well authenticate you by your voice. Of course it remains critical that this is done robustly to deepfakes and other fraud attacks
Authentic Leadership | Strategic Transformation | Financial Services | Risk Management | Customer Contact | Biometric Security
1 个月If the dream of Agentic AI is to become a reality for organizations and consumers, the agentic AI must have autonomy to act. A vital component of "autonomy to act" is to have the "permission and delegated authority to act". For most, this means that you know who your customer is, with a degree of confidence that aligns with your risk appetite vis-a-vis the service need to be fulfilled. If your natural voice and your natural dialogue is the interface - then voice based security will be fundamental to success. It may be natural for many to turn their attention away from traditional voice channels (as we have discussed), but conversely those who continue to invest in and perfect their natural voice security solutions are in fact building real muscle that will give them competitive edge in the future.
This was a great series, really well done!