GenAI lags traditional AI and ML for enterprise data backup
Enterprises are using AI and ML technologies with their backups for classification or security over recent generative AI tools like chatbots, industry analysts say.
Enterprise data backup software has lots of uses for AI and ML capabilities. But organizations are less enthused about much-hyped generative AI technologies.
More than 90% of organizations said that AI and machine learning features are important for their backup workloads. But they have concerns about the capabilities of GenAI, according to a recent report by TechTarget's Enterprise Strategy Group.
The Reinventing Backup and Recovery with AI and ML report, published in June, surveyed 375 of IT professionals from private and public sector organizations in North America, about how their organizations are adopting AI and ML in backup and recovery strategies.
These respondents considered data backup security, automated data classification with retrieval and automated backup and recovery to be the most beneficial use cases for implementing AI and ML.
GenAI is considered a positive technology for improving backup security among respondents with 47% strongly agreeing with the statement "generative AI will help us create an ideal backup and recovery process for our environment." However, that excitement is tempered by reservations around data quality and legal liabilities as well as a lack of interest in backing up data created by GenAI applications.
This temperance isn't unexpected, according to Jon Brown, an analyst at Enterprise Strategy Group and co-author of the study. Organizations have seen the value in using automation from AI and ML in their backup environments to lighten employee workloads and ease day-to-day operations.
GenAI has the potential to further expand backup management to more of the organization or assist experts during recovery operations. But the implementations need to meet the organization and regulatory requirements first, Brown said.
The recent CrowdStrike cloud service outage might have brought some of the dangers of automation into the public consciousness, Brown said. The benefits of automation technologies and the potential of GenAI will ensure that the adoption of both continues. However, organizations are wary of handing over too much responsibility to automation.
"It's the risk-reward we see in everything," he said. "The benefits are completely and utterly obvious, but we don't want to be that one guy at CrowdStrike who let it rip."
With GenAI, Brown added, data governance "gets even trickier" as accountability and fault for models as well as the data generated by said models has yet to be tested legally.
Survey respondents cited data backup security, automated data classification and retrieval as well as automated backup and recovery as the Top 3 use cases AI and ML where would have a positive impact.
Those surveyed indicated they'd likely use GenAI to develop recovery plan strategies, improve technical support and generate ransomware recovery plans and scenarios.
Backup vendors have already started compiling GenAI assistants and chatbots to enhance their offerings.
Lacking trust in GenAI
Organizations are interested in GenAI for data backup uses but are hesitant to adopt it widely, industry analysts agreed.
The trust placed in more established automation technologies isn't currently extended to GenAI capabilities offered by vendors, even with outlier events like CrowdStrike's error, according to Krista Case, an analyst at Futurum Group.
Those concerns and the relative infancy of GenAI in the enterprise will make enterprise IT hesitate at turning over more duties to automated GenAI tools, Case said.
"They have to learn to trust the tool," she said. "They have to be able to understand the use cases and understand the value they're going to achieve at the end of the day."
The challenge of adopting automation for GenAI backup tools also comes from the level of specificity in an organization's recovery environment, said Mike Small, an analyst at KuppingerCole.
"[Data recovery and backup is] extremely open to help from GenAI technology," Small said. "But like everything else, you have to teach it what the problem is. But everybody's problem is slightly different."
Establishing that trust in GenAI and ensuring its users are savvy enough to understand the ramifications of data storage and usage will remain a pressing challenge, said Max Mortillaro, co-founder of TechUnplugged. Backup teams need to understand how and where GenAI apps are creating data while helping the rest of the organization understand what's important to save.
"The question is, when are we going to give [GenAI] enough confidence for that to happen," Mortillaro said. "[You] want to protect [your] organization's data, but how does the [backup] team understand challenges beyond data protection?"
A GenAI copilot or assistant within most backup vendor platforms might not attain that level of capability. But Brown said proliferation remains inevitable now due to market hype and enterprise IT seeing the potential.
"You have to tack GenAI onto any AI product now. It's a user interface that's geared towards a less-sophisticated user of their product," Brown said. "It's important; it's incremental. But I don't consider them breakthroughs."
Questions surround data
Organizations might experiment with creating new GenAI co-pilots and chatbots, with about 30% of the ESG survey respondents saying they are running between 7 to 10 petabytes of enterprise data to train AI models.
Despite the amount of data these models are processing, however, only 65% said they regularly back up data created by these models. Those that are backing up the data said they're only backing up around 50% of the created data.
"We see AI as something that [needs protection for] both the source data and the data that comes out of your models," Brown said.
The benefits are completely and utterly obvious, but we don't want to be that one guy at CrowdStrike who let it rip.
Jon BrownAnalyst, Enterprise Strategy Group
Ignoring backups, either by accident or intent, might be a general reluctance among organizations to keep and maintain GenAI applications in the future, analysts said.
But IT vendors, such as IBM, see the opportunity in having more organizations use GenAI tools, requiring backup specialists to anticipate a data influx, Small said. Users outside IT will create code or applications specific to their needs with GenAI tools, he said, generating entirely new sets of data with specific needs. This would make backup teams use GenAI to quickly create specific backup services or protocols.
Organizations need to know where the data that GenAI applications depend on lives and where it's backed up, Small said.
"Not only [is backup software] protecting your conventional data, but it's also protecting your inferencing rules or data around those applications [so] you can continue functioning if something happens to them," he said.
Tim McCarthy is a news writer for TechTarget Editorial covering cloud and data storage.
