It seems like everyone and their mother is trying to jump on the AI bandwagon these days. But while you’re busy chatting about AI at your friend’s backyard barbecue, President Biden is in San Francisco meeting with AI experts to discuss the potential risks and regulations surrounding this buzzy technology.
Generative AI, the belle of the tech ball, has been on everyone’s lips since OpenAI’s ChatGPT made its grand entrance in late 2022. But are we letting the hype distract us from having important conversations about the safety of AI technology?
As Hal Daumé, a University of Maryland professor of computer science, points out, “There’s a lot of snake oil out there.”Essentially, companies are selling the dream of AI rather than focusing on what it can actually do.
Generative AI, which allows users to receive human-like responses from chatbots like ChatGPT, is a prime example of this. And while we’re all busy talking about how cool it is, discussions about its potential harms and ethical implications might be getting pushed to the wayside.
AI in our cereal boxes and toothpaste
It’s hard not to notice the AI craze. From emails to investor pitches, AI seems to be the golden ticket for grabbing attention. But as Brian Sozzi, Yahoo Finance executive editor, points out, there’s a glut of companies using AI as a shiny lure to reel in investors.
Erik Brynjolfsson, a senior fellow at Stanford University’s Institute for Human-centered Artificial Intelligence, explains this phenomenon: “People who don’t have the real goods will maybe have a little FOMO [fear of missing out] and try to join in the party.”
Take the education sector, for instance. Companies are all too eager to tout their AI-based solutions, but the real question is whether these solutions actually use AI and, if so, whether it’s genuinely beneficial.
The problem with this AI fever is that the constant buzz about AI products might be overshadowing the actual issues that come with the technology.
The double-edged sword of generative AI
Generative AI is undoubtedly impressive. It has the potential to revolutionize tasks like web searches, trip planning, and email writing. But it’s this same human-like quality that raises concerns about its ethical and safety implications.
And while companies are rushing to build their own generative AI models, they might not be considering the long-term costs and resource constraints. Naveen Rao, CEO and co-founder of MosaicML, points out that training generative AI models like MosaicML’s MPT-7B can cost as little as $200,000, but there are other costs to consider, such as keeping the AI up and running.
Rao also mentions the current GPU crunch, which is making it difficult for developers to secure the necessary computing power to build or run their models. He estimates that we’ll be living in a world of GPU shortages for at least two to five years.
Time for a reality check
It’s crucial that we don’t let the hype of AI blind us to its potential dangers and ethical implications. As we continue to develop and deploy AI technologies, we need to ensure that we’re doing so responsibly and with a focus on safety and transparency.
As for the companies that are using AI as a shiny lure, it’s high time they shift their focus from selling the dream to addressing the real challenges and possibilities of AI. And for the rest of us, let’s continue to have those important conversations about AI – even if it’s just at our friend’s backyard barbecue.
