Speculation at the end of last week was that OpenAI, the company behind ChatGPT, were set to launch a search engine. However, before this, Sam Altman, CEO of OpenAI, quashed the rumours before any such search engine launched.
Announced on the OpenAI Spring Update, the launch was ChatGTP-4o, the next evolution of voice assistance. Whilst still in its early stages, ROAST think what is emerging now is the 2nd iteration of voice assistants: the first-generation, Alexa, Hey Google, Siri, and Bixby, introduced users with the notion of ambient assistant access via smart speaker or your phone. And, yes – these assistants were a step in the right direction – however, expectations didn’t quite meet the perception of a digital assistant, a perception stemming from what is seen on TV, movies, and sci-fi.
OpenAI has now used its various AI technologies in addition to existing voice technology (speech-to-text and text-to-speech) to create what we can now think of as a smarter voice assistant which may have the chance to live up to those sci-fi expectations.
In the spring technology update on Monday 13 May 2024, OpenAI demoed GTP-4o being used as a live translator, solving maths problems, helping with code and all in a much more conversational and emotive manner, in comparison to previous voice assistant interactions. However, the key point here is that all is being driven by a GTP model, where previously 1st generation voice assistants, such as Alexa interactions would have been hard-coded by developers within voice apps / Alexa Skill.
This meant that if you tried to have an interaction and there wasn’t a voice app installed, your conversation would not go very far, whereas now your conversation could go anywhere and not be restricted by the marketplace/developers.
It was also interesting to see an extra input method, the camera. Previous voice assistants have been voice first as an input method, with screen-based interaction as an optional extra. In OpenAI’s demo, the assistant could open the screen to see what’s happening, with fun examples such as judging the outcome of a rock paper scissors game, viewing your device screen or using your device camera to see what is being shown by the user to provide help with tasks such as math equations.
A game of rock, paper, scissors with GTP as the judge.
Bringing this back to digital marketing, we had a couple of thoughts. A better voice assistant could start to shift some tasks which you would normally do on Google, to the assistant. For example, in the early stages of planning a holiday, the assistant could be more tailored to your needs and generate ideas. However, when it comes time to find flights or other bookings, Google would likely still be most users’ preference.
Another impact for digital marketing teams is more specifically for the teams to use these tools to improve their productivity and abilities to work. Asking the assistant to write a summary of performance by sharing a screenshot from Google Analytics, writing Python code to automate reporting or checking compliance rules on ad copy.
Uploading of data to create analysis and charts.
It’s not the search engine we were expecting, but another move in the direction of the assistant we’ve been expecting for a long time.
Read more over on OpenAI.
Interested in knowing more about how Voice and AI can be utilised in digital marketing? Reach out to our ROAST Labs team.
