The Touch
Death of the voice AI echochamber; birth of conversational AI convergence
There is no film that holds the special place, for kids of the 80’s like myself, than the 1986 Transformers animated movie.
The film featured an all-star team of 80’s talent, from Leonard Nimoy to Robert Stack to Monty Python’s Eric Idle to Casey Kasem to Weird Al to the last appearance of the legendary Orson Welles (as the nightmarish villain Unicron) coming mere days before his death.
The film is best known for taking the previous 65 episodes of the show, in which a grand total of zero characters died, and throwing that away to accommodate Hasbro execs who wanted to replace the old Transformers with new ones to sell more toys.
The early plot of the film involves a surprise attack by the villainous Decepticons, creating a moment where leader and beloved hero Optimus Prime realizes he either acts to save his comrades or everyone dies.
The resulting sequence of events causes his own demise, but not before we get Stan Bush’s memorable 80’s anthem “The Touch,” a song revisited and parodied and revisited again many times since then in popular culture:
Thus, a grand total of 20 minutes into the film and not advertised in any way in the film’s trailers, the filmmakers kill off Optimus Prime in a stirring, solemn scene that traumatized a generation for many years to come.
This scene is regarded as so traumatic and influential, filmmakers avoided death of main characters in films altogether for years after this.
Behold, cinematic history:
We don’t do very well with death, as a species, which might explain why the conversational AI landscape is having trouble coming to terms with what has taken place the last few months.
The body count the last 2-3 years in what I’d call the old guard of voice AI is high. Some of the casualties include:
the vast majority of the Amazon Alexa ecosystem of skills
100% of the Google Assistant “action” / third-party developer ecosystem
smart speaker sales, which are uniformly down and continuing to decline
numerous voice AI companies that were operating are now gone, with numerous executives having also moved on
Correspondingly, the previous echochamber of voice AI thoughts and narrative has been silenced, summarily replaced by a much larger community of leaders brought to us by the resonant, impactful, and seemingly-permanent arrival of ChatGPT.
With any death, something new rises to fill its place.
In this case: the birth of true convergence of conversational AI.
What does this mean?
Well, let’s first define conversational AI to be any semblance of AI (not interested in discussing what does or doesn’t constitute “AI,” just roll with me here) that involves voice, text, or chat as input. Algorithmic, pattern-matching computing involving natural language.
Next, let’s put the term “generative AI,” our current buzzword, in context: conversational AI is inherently generative in nature, in that a user provides a little bit of language input, and gets more than what they put in, often a lot more.
So “generative AI” is, by definition, subsumed entirely by the term conversational AI, with its only value, for this particular fleeting moment, remaining in marketing or PR contexts.
Now, with all that out of the way, let’s get to the main point:
If you are doing anything whatsoever with voice, you are inherently now working with chat and text. They are on your roadmap, whether you realize it or not.
And conversely, if you are working natively with text, voice is now something you must integrate.
Users of conversational AI - people interacting with technology using their natural language - will expect to be able to engage with you in their choice of spoken word or written word.
For as basic of a conclusion as this sounds like, you have to realize something: if you traveled back in time to the year 2019, no one would be thinking like this.
The voice AI ecosystem was squarely focused on reducing friction within that modality alone, while text and chat were not as harmonized with voice as they are today.
ChatGPT is simultaneously the Grim Reaper, delivering rapid death to the previous voice AI landscape and echochamber that it perpetuated, and change agent, creating something new to fill the void.
Convergence of all natural language modalities seamlessly into one consumer-expected approach.
There are far-reaching implications for this, one obviously one being that conversational AI companies have potential to be stretched thin delivering on the customer experience now expected by users of natural language-driven AI.
Which immediately forces smart companies to start to think about which industry verticals they truly belong in, and can win, versus which ones they play in simply as a vanity project.
It means the more focused you are even on one particular industry, the more likely you are to succeed.
It also means to participate in this emergent technology, the type of people who are your customer, the type of people who are your partner, and the type of people who are your competitor are now on the march, and the folks down in marketing have a tall task to try to keep up with who is who.
As a closing note, it’s helpful to note what happens historically when we see convergence within technology: the very next thing that happens is convergence yields to competitiveness.
You can be assured Amazon, Google, and Apple have very little interest in continuing the current path of OpenAI and Microsoft dominating daily AI headlines.
There’s no better time to be a startup than when the big players are competing. The clash of titans creates a lot of space down below, where smaller players can claim space and flourish and rise.
From death, we have new beginnings.
Until, at least, another massive new technology swoops in to rewrite everything again.
The formal press release on the Ethics & Integrity Charter for LLM-Based AI, a shared vision between the Open Voice Network/Linux Foundation and Project Voice, is available here, while the online petition can be signed here.
Registration and program information for the upcoming Project Voice 2023: The World of Conversational AI conference can be found here.