It’s easy to see why organizations are plowing billions into artificial intelligence (AI): The productivity gains and the chance to out-innovate competitors are just too tempting to ignore. But an irrational fear of running afoul of AI laws are paralyzing decision-making and preventing more companies from tapping the tech’s true potential.
I hear from quite a number of Fortune 500 companies that would like to integrate AI into their human resources practices, but are concerned they’ll trip a legal wire and prompt a government investigation.
If business leaders only look at the bad side, then the fear of AI is understandable. If the tech were weighed against the benefits, however, it has the opportunity to advance human ingenuity and innovation at unbelievable scale. Assigning the right level of risk/reward to AI is paramount so companies aren’t afraid to engage with it.
The first place to start is education. Those of us who create AI programs are not living up to our responsibility to educate others.We expect people to trust us. We have a duty to teach and tutor all who inquire, as obvious as the technology may be to the rest of us.
REGULATE AI, NOT INNOVATION
While lawmakers on Capitol Hill hammer out a bill to keep the United States competitive, chances of a deal in an election year are slim. A broad presidential executive order is currently the only federal directive aimed at managing AI’s safety and security risks and protecting privacy, civil rights, consumers, and workers.
In the absence of a federal statute, states are forging ahead on their own. California’s measure pulls back hard on the reins of the technology’s use in everything from housing to jobs. Colorado’s law applies to how AI is used in evaluating people for school applications, loans, and other things.
Likewise, other countries aren’t waiting around for the U.S. to act. The European Union’s law mostly hits the right notes. But in the totality of the regulatory environment, it is also concerning for businesses that have to deal with a hodgepodge of inter-territorial rules. What may be allowed in Germany might not be so in the Netherlands.
If other nations overregulate to the degree at which the EU has been doing for years, they will be unfriendly to business and innovation. To be clear, it is not because of one set of regulatory laws. EU AI in and of itself would not have an impact on innovation. The issue is over-regulation in the individual territories themselves. Even if they share a border, they don’t necessarily talk to one another. That’s a concern.
CORE AI VERSUS GENERATIVE AI
It’s only natural that business leaders with factories in different parts of the world are skittish about employing AI. But let’s set the record straight about what is and isn’t being regulated here.
There are two types of AI: core AI and generative AI. Core AI is traditional machine learning that has been around for decades. Then there’s generative AI, which has exploded in news coverage in recent years and been the subject of congressional hearings.
Both come with a certain level of risk, but core AI has been proven time and time again. It is used to calculate a person’s credit score, diagnose and prescribe treatment for an illness, and predict the weather. Most people don’t even think about it because it quietly does its job in the background.
GenAI, on the other hand, differs because it generates things—text, video, music, and world famous paintings. The inherent problem is that it uses publicly available data, which raises a host of thorny legal issues. Is it permissible to use someone else’s work and make it part of something new? For example, can a country music star’s voice be copied and used to make a new song without consent?
My guess is no. Even if something is copyrighted, unless it is a direct copy of someone’s work and is being sold as an original without crediting the original artist, it is most likely illegal. But existing copyright and trademark laws already cover that. New AI laws are not needed for those use cases.
What we do need is a law to notify viewers when photos and texts are generated by AI. A photo snapped on a phone already embeds loads of metadata—the date, time, and location where the photo was taken. Why should it be any different for AI-generated output? As long as someone understands where a piece of art originated from, there should be no issue. Yet, this is where the fear is coming from—no one really knows.
Harmful use of AI should also be regulated. Using AI to seek vengeance or hurt someone should be illegal. So we need to regulate the bookends—educating so that we build awareness while also preventing known harm.
Until then, it’s understandable when business leaders balk and say, “I’m not going to risk my company’s reputation using these tools.”
Here’s how business leaders can reduce their risk.
- Analyze your current technology ecosystem, because there’s a good chance you are already using AI and may not even know it.
- Understand the difference between core AI and GenAI, because that will help executives understand how to introduce AI at scale and implement it safely. Start with specific use cases. For recruiters, use AI to generate emails to job applicants or analyze job descriptions. Take a low-value human task and turn it into a high-value intelligent experience.
- Talk with experts in the field. They can help make your business case with all sorts of ROI proof points. It is valid to ask what’s the risk/reward with AI? Do the homework.
If none of these three steps eases concerns at your organization, and there’s still hesitation to bring AI into the fold, then maybe it’s time to move on to an employer that gets it. Because that skittish company is at serious risk of being swallowed whole by a competitor that is already light years ahead using AI.