Reliable cyber infrastructure key to AI readiness, says ManageEngine CEO

JAKARTA — Cyber infrastructures need to be reliable to support emerging technologies as scaling amid increasing demand for artificial intelligence (AI), according to the top official of ManageEngine, the enterprise IT management software division of Zoho Corp.

“Sometimes AI is magical, but it is too good to be true. It means it also comes with a lot of caveats. It’s not going to stay like this forever because something like token cost is only something we have discovered in the last few years,” ManageEngine Chief Executive Officer (CEO) Rajesh Ganesan said in a speech at the ManageEngine User Conference for Southeast Asia held here on Tuesday. “We are going to keep finding these caveats one after another.”

He said data centers require a lot of power, and increasing use of AI services will drive this up even more.

With most companies globally still catching up and figuring out how to scale their AI capabilities, Mr. Ganesan said the need to ensure they have resilient digital infrastructure and address bottlenecks related to digital autonomy.

“Almost 80% of the corporations in the world are still AI curious and still only running a pilot project. And then there are some companies that are AI-ready. They know what problems they have. They know how to use the software using the app. They are ready to leverage it. There are very few companies across the world today that are AI-driven.”

Organizations’ infrastructure reliability is being tested by the scale and complexity of emerging technologies, while operational resilience is weighed down by fragile systems and the lack of unified models, he said.

Companies also need to enhance data quality and sovereignty, he added.

“The context starts with the reliability of the digital technology. Do you have a clear understanding? Can your instrument scale up to what autonomous areas you want? Can you harness and assess the portion of the data quality that you have? Do you have data science? Do you have absolute confidence to say where all your data is? Do you have data entry points? These are uncomfortable topics that need to be asked before you even think about leveling up your AI,” Mr. Ganesan said.

“Because if your data is not good, regardless of the power of a model, you are not going to succeed.”

Companies also face power and compute constraints as their foundations are limiting speed and autonomy, he added.

Governance and compliance, as well as security issues, are also concerns commonly faced by companies looking to expand their AI services. — Aaron Michael C. Sy