Echoes of Us: Too Human to Ignore - The AI Voice Dilemma
Google NotebookLM - When AI voices stir real feelings - we need to take notice
Alan Watts said:
The only way to make sense of change is to plunge into it, move with it, and join the dance
Wise words? Perhaps.
But it depends on what is changing and how fast it's changing, right?
It's hard to plunge into a change that disturbs us.
Each of us has areas where we are open to change, and other places where we find change hard.
I find changing my morning routine hard.
I make my coffee in the same Italian Moka pot and take my vitamin pills while it brews.
Then I sit down with my iPad, reading the latest news on Substack, while I sip my coffee.
Before I do anything else, every morning, I love it.
When I've been in situations when that's been disrupted, or even worse, I don't have any coffee, I'm just not the same, I'm disturbed.
Know what I mean?
Where in your life do you find change hard?
One of the areas many people find change hard is the rapid pace of AI change.
Every day, there's something new in AI. If you find it overwhelming, I do too.
Let's be honest again, there are far more people worried about AI, often quietly, than there are excited about it.
Many of those worried about AI, just want it to go away, or avoid it.
If you feel that, I get that.
I also believe, when you can understand what AI is better, and what it means for you, I know you'll be able to face these rapid changes with more confidence.
That's what I want to help you achieve.
How AI Voices Are Becoming More Human
There was an important AI development last week that was not well reported on, which I think you need to know about, however, you feel about AI.
Google released an AI that creates hyper-realistic voices that are uncannily real.
If you think you know how AI sounds, clunky, like a robot - think again.
I’ve been an AI consultant & AI researcher, I'm in touch with the progress of AI, and I'm not easily surprised or astonished by the latest AI developments.
This one blew me away, I'm astounded.
You need to know about this, because one way or another you're going to run into this sooner or later, and I want you to be prepared for it.
When you can't tell the difference between what's real or artificial, between what is human and not human, I think that matters, and you need to know.
So, let me fill you in...
Around 2 weeks ago Google released an update for an experimental product NotebookLM, with no announcement, no press release - nothing.
Before the update, NotebookLM was an ok AI tool, and I mean as that someone very familiar with AI and tools like ChatGPT.
As I said, I'm not easily impressed.
AI-Generated Content: The Future of Podcasting and Content Creation
NotebookLM is basically like ChatGPT, but specialised for research and study, you can upload documents, webpages etc, then ask questions, summarise, that kind of thing.
So what was this update?
It was something innocently called 'Audio overview' - so what does it do?
It turns uploaded resources and documents into a podcast.
What that means is, that it creates an audio file from your documents, of two hyper-realistic human sounding people, discussing and talking about your references and documents.
Probably the best way to help you start to understand, is to show you.
I've created several podcast episodes, using different articles of mine as a basis for the podcast discussion, here's one, have a listen.
So, now you know why I wrote this post...
What do you feel listening to these voices...surprise? intrigue? concern?
All very understandable if you do.
When I first heard these, I was shocked.
It's not just because these voices sound like real people.
I think that's an understatement.
It's the little things in their voices that hit me the hardest, not what they said, but how they said it.
The intonation, the rhythm, the joy, dare I say it, I felt the emotion in what they were saying, their excitement, their scepticism.
Did you feel something too?
When I say I felt it, I mean I felt things from these voices speaking, I laughed at many of their jokes, and I really pondered their more thoughtful questions.
Strange as it may sound, these AI voices really touched me. I warmed to them and enjoyed listening to them.
Isn’t that strange?
I think it's very likely, that you felt something too listening to these voices.
After careful thought, I've decided to experiment with some of these AI-generated discussions on my podcast in addition to the episodes I speak.
But why?
The Impact of AI on Human Emotions: Strolling through and beyond the Uncanny Valley
If you are concerned about the potential for deception, because these voices are so realistic, that's a very valid concern and also why I have highlighted they are AI-generated in the podcast show notes I publish.
I suspect many people will have strong reactions to hearing those AI voices that sound too human,
Some might be delighted and amazed, others might feel fearful or even angry.
These are all understandable and valid feelings that are important to acknowledge.
Feelings are what make us human, after all, and even when they are difficult, you are stronger for feeling them than ignoring them.
An academic called Masahiro Mori came up with a theory for this range of strong feelings about human-like AI we have, that he called the Uncanny Valley which he describes as:
Uncanny valley, is a theorised relation between the human likeness of an object and a viewer’s affinity toward it. The hypothesis originated in a 1970 essay by Japanese roboticist Masahiro Mori, in which he proposed that as human likeness increases in an object’s design, so does one’s affinity for the object—but only to a certain point. When the likeness nears total accuracy, affinity drops dramatically and is replaced by a feeling of eeriness or uncanniness. Affinity then rises again when true human likeness—indicating a living person—is reached. This sudden decrease and increase caused by the feeling of uncanniness creates a “valley” in the level of affinity.
This is what the uncanny valley looks like as a graph:
I've highlighted where I believe NotebookLM feels best placed to me on this graph in terms of how close to sounding like real humans it is.
I feel the affinity for these voices has gone past the point where they sound weird, beyond the uncanny valley, getting very close to sounding very human and, even likeable.
What do you feel?
My main aim is to help you better understand and navigate AI.
This is the key reason I've decided to experiment with using these AI-generated conversations for some of my podcast episodes alongside my human ones.
Here are my thoughts on why I did this, and how this can help you.
How Familiarity Can Help You Navigate The Rapid Technological Changes of AI
Explaining AI in a way that's easy to understand can be challenging at times.
Showing you, rather than telling you, I believe can sometimes be a better way to help you understand what's possible with AI.
I'm also a great believer in the dialectic method of acquiring knowledge, that is, acquiring knowledge using dialogue between people, as opposed to gaining knowledge by reading for example.
I think it's a great way to learn things that work well for many people, which is exactly why I created my podcast, which includes discussions between me and guests.
So using these very engaging AI generated conversations about the topics raised in my content, I think could be a useful way to help you better understand what AI is and its potential impact, on many levels.
But there's more.
I know the articles I have written very well of course. I have now listened to many of these AI-generated podcast discussions of them.
What I can see is this, they are not only discussing the points raised in my posts, but they are also making new points & insights and original contributions beyond those I made, that are highly relevant and useful to the issues I raised.
This was one of the most surprising things about these AI-generated discussions for me.
In my view, they are not just talking about the points in my articles, but adding their unique insights, making connections with other issues and points, in a highly insightful & intelligent way.
I found it a joy to listen to these AI conversations not only because of their engaging realistic voices full of emotion, personality and expression, but because I also heard new and original points of view beyond my own that were excellent contributions to these debates.
I actually learned more about my topics, by listening to these AI discussions.
The irony is not lost on me, that AI discussions might help you get AI better.
So I'd also like you to be able to experience these new and original AI-generated insights too, when thinking about the issues raised in my posts.
Knowing Whats AI and Whats Human
If you are feeling anxious about these AI developments, I get that, and this knowledge can I believe help you too.
For now, at least, these AI voices generated by NotebookLM sound the same, you can't change them.
This also means this is how you can know when voices are AI-generated using this NotebookLM AI tool when you hear any audio content.
I'm sure you won't be surprised to know, that many people are experimenting already with creating audio content using this tool.
But at least for now, you'll be able to know if they are AI-generated because it will be the same voices that sound like these.
If you believed AI-generated voices would always sound clunky and like robots and never sound human, I hope you now better understand how our assumptions about what AI needs to stay flexible and open to new possibilities.
You can also learn from this that AI is rapidly changing, what it can and cannot do is not staying the same. Thats a valuable thing for you to understand.
AI and Human Connection: Navigating the Uncanny Valley of AI
I know, that the more you understand about what AI can and cannot do, you'll be better prepared for the future.
The best thing we can do in response to AI sounding more and more human, is to better understand it by becoming more familiar with it first.
Could the internet be flooded with AI generated content where we can’t tell whats human or AI?
Or, could this be a useful tool that can help content creators create engaging and helpful content?
Before we can figure that out, we must first better understand what it is, what it can do, and not do. We can only really do that, by seeing how it works.
This is why I will be creating more podcast episodes using this AI tool, by doing so I think these will help you better understand AI's potential in a way that's you can easily experience and understand.
I'd love to know how you feel about these AI podcasts, do they frighten you, or excite you? Do they help you better understand the topics discussed?
Do they sound identical to humans to you? or a bit different?
The rapid technological change that AI is bringing, especially as the lines between what is human and AI get more blurred, can bring up a range of different feelings in us, some of which might be difficult feelings.
How should we use this technology? Should it be regulated?
You might have many questions, and I totally get that.
It can be helpful to discuss these with others in a safe space.
That's why I created as part of my paid membership, an exclusive paid member chat where you can discuss these issues, your concerns, worries, hopes and questions with others like you and me.
When you can understand what AI is better, how you feel about it, and what you can do about it, I know you'll be able to better face these rapid changes with more understanding & confidence, and that's what I want to help you achieve.
I’d love to help you better understand and navigate these AI changes in our exclusive member chat, to inform and empower you.
Find out more about paid membership here.