Cloud Hosting

AI has crossed over from being a tech buzzword into a household name, but consumers and investors should be circumspect when thinking about the AI hype, said Duke University professor Sultan Meghji, who previously served as the FDIC's chief innovation officer (video above).

"So much of AI right now is just driven by marketing teams and not actually by technologies, so I think it's not surprising at all that we're talking about the hype cycle," he told Yahoo Finance Live. "So much has been announced, there are so many glittering logos and great press releases, arguments on social media, things like that but we haven't actually seen a lot of work being done."

It's not that Meghji doesn't believe AI will be transformative he does, he just thinks the biggest changes AI brings about are going to be less headline worthy.

"The current generation of AI is really going [to] start having impacts over the next few years, where it does things like streamlining back-off processes, but that's a completely separate set of activities from what gets covered by the news," he said.

Some tech giants, especially Alphabet (GOOG, GOOGL) and Microsoft (MSFT), have leapt headfirst into the AI craze, but others like Apple (AAPL) have held back a bit.

"I'm not surprised at all that they've taken a slightly slower approach, a more engineering-centric approach, than many of these other organizations," said Meghji. "As one of the few trillion-dollar tech companies out there, have a massive consumer base ... If, all of a sudden, you have to add 100 million users to an AI system, you have to have a fair amount of infrastructure behind that and it's possible that they just don't have it."

An Apple store employee stands inside the store in New York on Feb. 5, 2021. (AP Photo/Mark Lennihan.)

Apple's slowed approach has proven to be the exception rather than the rule, and much of the conversation that's following AI right now fixates on the most extreme scenarios, both good and bad. However, the key technological challenges that will lead to us getting AI right or wrong are considerably more mundane. For example, one of the central problems that Meghji expects AI to face moving forward, especially in the case of large language models (LLMs) like ChatGPT, is appropriately curating what data it does, and doesn't, train on.

Story continues

"So many of these systems are just based on fundamentally learning from what's openly available on the internet and, let's be honest, a lot of what's out there on the internet isn't that great," said Meghji. "If you run out of data to train, because you've looked at the entire internet, you can end up looking at data that's been generated by your own system in essence, drinking from the same well that you're feeding into."

He added: "In AI, you have to be really careful with the data that you use to train it and, at some point, you'll stop having positive returns. It'll stop getting smarter, so you need to stop and take a step back."

Allie Garfinkle is a Senior Tech Reporter at Yahoo Finance. Follow her on Twitter at @agarfinks and on LinkedIn.

Click here for the latest trending stock tickers of the Yahoo Finance platform.

Read the latest financial and business news from Yahoo Finance.

Download the Yahoo Finance app for Apple or Android.

Follow Yahoo Finance on Twitter, Facebook, Instagram, LinkedIn, and YouTube.

See the article here:

AI hype is 'driven by marketing' right now, says former FDIC chief innovation officer - Yahoo Finance