Thanks to the “GPT effect,” you may see AI chatbots everywhere

ChatGPT’s success could lead to more AI bots teaching how people use the internet, tech experts said. Getty/contributor NurPhoto

• Microsoft’s new Bing is powered by technology from OpenAI, whose megahit bot ChatGPT is redesigning search.

• Google announced its own AI bot, Bard, which is still being tested.

• ChatGPT’s success could lead to more AI bots teaching how people use the internet, technology experts said.

Large language model tools aren’t new, but the instant popularity of OpenAI’s ChatGPT has put Microsoft against Google in the burgeoning online search battles and could even change the face of the internet.

Microsoft’s new Bing chatbot, powered by OpenAI’s technology as part of its “multibillion-dollar” partnership, is still working through apparent glitches. (Bing’s AI chatbot has already made a name for itself for its animated and bizarre responses among users who have had a chance to test it).

But the hype still seems to be going strong – Microsoft executive Yusuf Mehdi tweeted this week that “several million” are on a waiting list to try the bot and that it’s being tested in “169 countries.”

Google, meanwhile, appears to be working on launching its own Bard AI chatbot, with CEO Sundar Pichai urging employees to work on getting the bot operational.

The technology promises to transform search from an everyday query that returns a page of results. For example, Insider’s own testing of the new Bing revealed that it’s an interactive process, in which the bot can sometimes act like an assistant to quickly browse the web to provide specific travel and sightseeing recommendations, or create daily plans .

But this AI-assistant-type conversational model could be applied to other activities consumers do online, like shopping or managing their budgets or tracking their movement, said Prasanna Arikala, chief technology officer at Orlando-based Kore.ai .

The challenge, as OpenAI’s own leadership has acknowledged, will be ensuring the bots are providing reliable information, he said.

“It will definitely change the way we live our lives,” Arikala told Insider. “But then it’s very important to understand where the line is, what to believe and how to responsibly consume that information.”

ChatGPT itself can also help train other AI models, said Cameron Turner, VP of data science at Kin + Carta, a Chicago-based consulting firm that advises companies on how to effectively use data to develop their business. Turner also previously led a key data-focused team at Microsoft called the Telemetry Group.

Turner said ChatGPT could help produce organized data to turn information into fodder that can train other AI models for more specialized applications. For example, AI technology is being developed to support health research, combat cyberattacks, and improve supply chain logistics.

Using AI in this way can help “if you want to experiment in a cheap way, without making a commitment, without making large-scale data purchases just to test the hypothesis,” he said.

And it’s not just the internet. Tech experts also said they hope OpenAI’s evolving language processing tool can improve other types of cutting-edge technologies that people interact with. For example, it could help robots learn to communicate, said Tesca Fitzgerald, an assistant professor at Yale University who teaches a class on artificial intelligence and does research in robotics.

Robots are taught and programmed to move, recognize their environment, and navigate their environment, each of which requires specialized expertise, Fitzgerald said. OpenAI’s GPT technology can be used to help robots “talk” to humans who control them, she said.

“We could use GPT to teach robots how to break down a problem into simpler steps to perform tasks,” Fitzgerald said.

Read the original article on Business Insider

Source