OpenAI kicked off a technological revolution when it released its ChatGPT artificial intelligence-powered chatbot into the world late last year. Generative AI looks like it will be a game-changing technology. With a simple prompt, AI models can generate high-quality text, create images, render videos, write code, and answer questions. Trained on mountains of data, the latest AI models are incredibly impressive.
But are those models safe? The Federal Trade Commission isn't so sure. The agency, tasked with protecting consumers, is investigating OpenAI for potentially harming consumers. The FTC is looking into OpenAI's data collection practices and whether its models produce false information on individuals. Speaking to the House Judiciary Committee last week, FTC Chair Lina Khan cited reports of people's "sensitive information" showing up in results, according to The New York Times.
OpenAI isn't a public company and can't be directly invested in, but the problem of opaqueness extends throughout the AI industry. Large language models like the one that powers ChatGPT can sometimes produce false, biased, and harmful results. These models are trained on enormous quantities of text. If any of that training text is toxic, the model will inevitably spit out toxic results on occasion.
How to safely bet on AI
One way to bet on AI without being exposed to the risks that a particular AI model will come under fire is to invest in a picks-and-shovels company such as Nvidia (NVDA 6.18%), whose data center GPUs are used to train AI models, and they've become the de facto standard thanks to their incredible computational horsepower and Nividia's yearslong effort to build a software ecosystem.
The downside of investing in Nvidia comes down to price. Nvidia is now valued at over $1.1 trillion, which is nearly 30 times the company's expected revenue this year. This is a valuation that leaves absolutely no room for error. If Nvidia doesn't grow as quickly as expected, perhaps because competition steals away some market share, it could be lights out for the stock.
Another good way to bet on AI is to invest in a company that's putting AI governance and safety first. International Business Machines (IBM -1.05%) has been a player in the AI industry for years, but its latest product could help solve some thorny problems for organizations looking to incorporate advanced AI models without the risk of those models producing problematic results.
IBM is in the process of rolling out its watsonx platform. The platform is composed of three parts, two of which are already available. The first component, watsonx.ai, enables enterprises to train, validate, tune, and deploy AI models. The second component, watsonx.data, provides a data store for the vast quantities of data required for AI workloads. And the third component, which is set to become available in October, is watsonx.governance.
For any large organization, deploying an AI model comes with risks. An AI model used for customer service may need access to customer data, for example, and that AI model absolutely can't be allowed to leak personal information. In the case of an AI model that provides health recommendations, the chance of a harmful recommendation needs to be eliminated. Using watsonx.governance, an organization can ensure their AI deployments aren't causing harm or violating regulatory requirements.
IBM's watsonx platform solves a variety of problems for big companies that want to use AI but don't have the appetite for taking big risks. With AI technology coming under government scrutiny, IBM's bet on safer AI looks particularly promising. And with the stock trading at a beaten-down valuation, it's an inexpensive way to bet on AI.
