AI: Data Management’s Catch-22

Simon Jelley, General Manager for SaaS Protection, Endpoint and Backup Exec at Veritas Technologies.

Exactly 160 years ago, novelist and critic Samuel Butler famously put forth that Darwinian evolution might apply to machines and that they could one day become conscious. Nearly 75 years ago, legendary mathematician Alan Turing created his well-known Turing test to measure machine intelligence. Twenty years ago, NASA’s Spirit and Opportunity rovers were making their way across the Martian surface, autonomously avoiding obstacles with the help of AI.

And now, OpenAI has unleashed ChatGPT. With a simple online or mobile app chat box user interface available to even the technologically novice among us, it’s perhaps the most advanced and accessible AI ever. Need an essay on quantum computing? Ask ChatGPT. Need to know how to arrange flowers for a bouquet? Ask ChatGPT. Need to know how to best connect with your church’s youth? Ask ChatGPT. (Yes, that last one is from personal experience as a member of my church’s pastoral council. It did a pretty good job!)

ChatGPT, along with contemporaries such as Google Bard, has ignited an unprecedented furor over AI. This reminded me that just over a year ago, I wrote an article that focused primarily on extolling the benefits of AI as it applies to autonomous data management.

The Challenges Presented By AI

But there is another side to the story. After all, Samuel Butler’s suggestion that Darwinian evolution may apply to machines ultimately led him to conclude that machines might eventually supplant humanity. And more recently, some of our time’s greatest minds have advocated pausing AI development until society can implement a set of safety protocols.

Indeed, AI is a double-edged sword. And it’s data management’s catch-22—just as it has tremendous benefits, it may also create potentially significant problems. For example:

• Cybercrime

It’s not out of the realm of possibility that AI chatbots like ChatGPT and the technology behind them could usher in a golden age of cybercrime where even the least technical criminals turn to pilfer data for profit. Need ransomware? Just ask. Or advanced cybercriminals may be able to use these tools to build even more innovative malware and attack vectors. Need a more sophisticated phishing scheme? Just ask.

• Volume, Velocity And Variety Of Data

I also previously wrote about the three Vs of data: volume, the total amount of data your company is creating; velocity, the rate at which your company creates that data; and variety, the number of formats that data comes in. When you strip everything else away, AI is really just absorbing existing data, “thinking” about that data, and then outputting even more data faster and in a greater variety of formats.

Thus, AI can also compound the challenge of managing today’s already vast amounts of data spread across the on-premises and multi-cloud infrastructures enterprises are dealing with, which has both security and cost implications. And with the world’s data—which is stored in largely fossil fuel-powered data centers—already generating as much carbon waste as the entire airline industry, the potentially overwhelming amount of unneeded AI-generated data could also have a significant environmental impact.

• Data Compliance And Governance

In addition, AI tools like ChatGPT create data compliance and governance challenges. Employees are turning to them to streamline their jobs. While this may improve productivity, there’s enormous risk when employees share confidential information, like regulatory documents, earnings reports or personally identifiable information, with chatbots. The AI may even use that information as part of its machine learning process to inform answers it provides to others.

Overcoming Some Of AI’s Challenges With The Help Of…AI

Make no mistake, though: I’m optimistic about the future of AI and the impact it may have on our lives, especially data management in an increasingly multi-cloud world. In fact, I believe AI-driven data management that does what IT teams can’t or don’t have the time to do is the answer to many of the very challenges the broader application of AI creates. Here are just a couple of the ways that AI-powered data management may address some of the potential problems AI itself could create.

• Dynamic Cyber Resiliency

The threat landscape was evolving at an alarming pace even before generative AI tools entered the scene, but now, it could accelerate to unimaginable levels. The solution may be to fight fire with fire. Adopting AI-driven anomaly detection and other similar security measures as part of a comprehensive data management strategy may help companies protect against the effects of the ever-evolving threat landscape.

• Data Services That Self-Provision, Self-Optimize And Self-Heal

Leaders in data management need to continue developing services that can autonomously self-provision, self-heal and self-optimize—including advanced autonomous deduplication. This can help account for the vast amounts of data in today’s multi-cloud environments, which is surely only going to increase as generative AI increases the volume, velocity and variety of data being created. As I stated previously: “In practice, this will look like autonomous provisioning of data protection policies when new services and users are deployed, and autonomous monitoring and rollout of new policies that match the observed usage of a company’s data—again, all with no human decision making needed.”

• Standardized Regulations

AI is already playing a role in advanced data compliance and governance solutions, such as staying ahead of potential compliance issues in an unprecedented number of business communication platforms, and it could eventually play a role in helping to prevent the leakage of data through generative AI tools. However, first, we must accept, advocate for and implement regulatory limitations aimed at keeping potentially sensitive information out of these tools. Until these regulations are in place, business leaders can join a growing list of companies implementing their own rules to prevent data compliance issues from arising as a result of employees using generative AI tools.

In summary, for all of their benefits, generative AI tools also come with unique challenges. But AI itself, along with adequate regulatory limitations, can help offset those challenges.

Forbes Business Council is the foremost growth and networking organization for business owners and leaders. Do I qualify?