Repeat after me: “This is the least capable, most expensive Generative AI landscape we will ever have.”
This maxim is core to every future development in AI: every system from here on out will be at least as capable as today’s systems, at least as fast as today’s systems, and no more expensive than today’s systems. The future of Generative AI is better, faster, cheaper than what we have today.
I bring this up because OpenAI made an announcement at their recent Dev Day that will reverberate through throughout the job market, across the world, over the next few years. And it’s only going to get more powerful.
Let’s dig in.
What is OpenAI’s Advanced Voice Mode System?
Earlier this year, OpenAI demonstrated a new ChatGPT system called Advanced Voice Mode that was astonishing both for it’s power and realism. It could answer and ask questions, modulate tone just like a person, make self-effacing jokes and laugh at itself, and even sing (badly). It had all of the intelligence of other ChatGPT models, but could understand your speech, tone, and emotion, and respond in a charming, realistic voice. In fact, it reminded viewers so much of the movie “Her” that Scarlett Johannson (the voice of the AI in the movie) thought OpenAI might have ripped her off (they didn’t).
Last month, this capability finally rolled out to paid users of OpenAI’s ChatGPT via the Android or iOS app. The length of the conversation was limited to an hour, but fully delivered on the promise of the earlier demo. While this was great for conversation, brainstorming, and note-taking for individuals, it was still limited to operation within the ChatGPT app and for a limited period of time.
At Dev Day on October 1st, OpenAI announced the release of Realtime API, which enables interactions like Advanced Voice Mode via API. This offers two main benefits. The first is that this incredible system can now be used outside of the ChatGPT app and integrated into any other app for realtime voice chat. The other is that these interactions can now be scaled effectively infinitely – no limitations on interaction length or the number of people that can be served (as long you pay for it).
All Tier 1 Support Will Become AI
We have reported before about the huge potential for Generative AI to function as Tier 1 support. This role is inherently an expensive one for businesses – very labor intensive due to the sheer volume of requests, but performing relatively simple, repetitive work for customers. AI was already making significant headway in supporting or replacing this role via text functionality alone, as documented in our Klarna case study among others.
With the human-like voice-to-voice (voice input from the user, voice output from the AI) functionality and intelligence of OpenAI’s new system, all of the pieces are available to offer fully automated Tier 1 support services through either text or voice. The system will scale as large or small as needed to match call volume, and businesses are only charged for exactly what they use. Staffing, infrastructure, and IT costs all disappear. Any knowledge updates can be pushed immediately to all agents, perfectly and instantly. And the overall cost for this system lands around at $9/hour/agent – competitive with off-shore call centers, without the associated hassle or reputational costs.
And – repeat after me – this will be the least capable, most expensive version of this system we will ever see. It’s only getting cheaper, faster, and better from here.
Huge Potential For Other Applications As Well
In their press release, OpenAI specifically highlighted other interesting applications, including language learning (did I mention the system was multilingual?) and education. This could also be easily extended to other common business functions, such as business development, receptionists (camera + real time voice API?), logistics, accounts receivable – OpenAI even demonstrated a version of this system placing an automated order for 400 strawberries (a subtle nod to their recent “thinking” AI, codenamed Strawberry). Any role which requires a high-volume of human interaction, but can be managed through a set of clear rules, could leverage this sort of system.
As with many other generative AI systems, the main speed bump right now is integration – users still have to make the code that connects this system, their data, and their users. However, OpenAI isn’t the only game in town, and other major competitors, such as Google and Anthropic, have shown a knack for offering their own competing AI in user-friendly, nearly off-the-shelf formats. Any number of startups will also likely rush to offer their own customized services to address every nook and cranny in industry. OpenAI’s announcement is just the beginning – within two years, this will functionality will likely be everywhere.
Generative AI has been gently working it’s way into businesses for the last two years, slowly establishing itself within a few key areas. Within this latest announcement, OpenAI has kicked down the door, and a trickle of applications will become a torrent – saving money and adding business value everywhere it reaches.
Become your company’s AI expert in under 30 minutes a month by signing up for the Executive Summary newsletter by AI For Business.
If you liked this post, have future posts delivered straight to your inbox by subscribing to the AI For Business newsletter. Thank you!