Exactly three weeks ago, OpenAI released ChatGPT.
Since then, it’s been nearly impossible to keep up with both the hyped excitement and hype brow frowning worries around use cases for the text-generating chatbot ranging from the fun (writing limericks and rap lyrics) and the smart (writing prompts for text-to-image generators like DALL-E and Stable Diffusion) to the dangerous (threat actors using it to generate phishing emails) and the groundbreaking (can Google’s full search model [subscription required] scammed?).
Is it possible to compare this moment in the evolution of generative AI for another technological development? According to Forrester Research AI/ML analyst Rowan Curran, yes.
“The only thing I’ve been able to compare it to is the release of the iPhone,” he told VentureBeat. Apple’s iPhone wasn’t the first smartphone, but it swept the competition with its touchscreen, ease of use, and introduction of apps that put an entire computing experience in our pockets. The release of the original iPhone in January 2007, followed by the launch of the App Store in July 2008, ushered in a period of historic technological change, Curran explained — as the general public discovered that there was a whole universe of creativity and applications with which to could work.
It made people aware “that you could have this portable computer that basically looks like [having] a Star Trek tricorder in our hand — this thing with tons of sensors and capabilities,” he said.
ChatGPT, like the iPhone, is changing the public consciousness
ChatGPT is also changing the public awareness of what is possible. But what’s happening now goes even further than that, Curran noted.
“I think what’s really unique here is that we have a technology that’s useful today, that’s advancing very quickly and that we’re all learning about in real time — both in terms of how to use it and how to avoid using it . in a negative way,” he said.
ChatGPT’s release and adoption cycle was also unique, he added. “There were a million users in the first few days — even if we assume a quarter of that is double, that’s still hundreds of thousands of human brains suddenly playing with this technology, which is very different from any other way we’ve released technology and adopted,” he said.
Was this a responsible way to release ChatGPT?
While some have criticized the way OpenAI released ChatGPT – venture capitalist, economist and MIT fellow Paul Kedrosky recently tweeted “[S]hame to OpenAI for launching this atomic bomb without restrictions in an unprepared society” – Curran insists it was “probably one of the most responsible ways they could have introduced this to the public.
OpenAI’s approach of iterating ChatGPT and showing it to people step by step is “a really good way to get people used to this, because otherwise this would all be happening behind closed doors at a large corporation,” he said, pointing to that even for those who pay attention and weren’t daunted by ChatGPT’s capabilities, progress is moving at a remarkable pace.
“If the public went straight to what comes after ChatGPT, people would go crazy when it came out,” he said. “I think OpenAI is trying to avoid culture shock with what they’re creating.”
Potential for seismic change in the enterprise
Just as the iPhone and apps eventually revolutionized all areas of the business — from software development and social media to customer service and marketing — Curran said he thinks ChatGPT and other generative AI tools could drive a “seismic change” in the enterprise. cause. in 2023, if companies and suppliers are aware of how they apply the technology.
“If we can avoid big, negative press events on this in the near term, I think the acceptance will be pretty deep because the appetite is very high right now,” he said. “You can see the ease with which people are already integrating [generative AI] in existing work systems, with a bottom-up approach — you can see this, for example, with Shutterstock, which integrated DALL-E two months ago, and now Microsoft has a beta access product called Designer, which is basically a text-to-image generator integrated with PowerPoint.”
Implementing best practices remains essential
And whether it’s ChatGPT or other generative AI capabilities, implementing best practices is still essential, Curran said.
“I think we’re still collectively figuring out what the exact best practices are, but there’s no reason not to continue implementing best practices around understanding your vendor solutions – if you’re getting a large language model through a supplier, which model do they use? What was the basic training data? What is the fine-tuning of the training data? How do they check this model?”
In the past, he added, companies have been set on fire by new technologies. “We never seem to really learn that when new technology comes along, we have to be thoughtful about its adoption,” he said. “But this time, because there are so many opportunities for people to get involved at a grassroots level, we can actually get people to step in and say, okay, I want to participate in this governance process.”