IBM doubles down on AI and hybrid cloud

IBM is beefing up its core technology capabilities with a slew of new offerings designed to make it easier for enterprises to leverage artificial intelligence (AI) and hybrid cloud.

Speaking at a media briefing before the start of the Think 2021 virtual conference, IBM president Jim Whitehurst claimed that the company has made “tremendous progress” in delivering on its vision of hybrid cloud and AI.

For one, Whitehurst said IBM is addressing a key challenge with AI implementations – having data required for AI modelling spread across an organisation, along with the lack of separation between data and application logic.

To that, IBM has developed AutoSQL, a universal query engine that is part of the intelligent data fabric capabilities of the next generation of IBM Cloud Pak for Data.

“It basically leverages AI to automate access integration and management of data without ever having to move it,” Whitehurst said. “It has elements of data virtualisation and it solves what I would argue is the most critical pain point as customers look to gain insights from their data.”

At the same time, IBM is also extending the use of AI to orchestrate business processes that span multiple applications through Watson Orchestrate.

Whitehurst said Watson Orchestrate uses natural language processing – with no programming needed – to connect applications like Salesforce, SAP and Workday, to data and tools such as calendars and e-mail.

“It’s based on organisational knowledge and prior interactions, and it can select the skills needed to make connections among applications,” he added, noting that professionals can use Watson Orchestrate to schedule meetings and even perform mission-critical tasks like preparing proposals or business plans.

On the hybrid cloud front, IBM is adding its Mono2Micro service to WebSphere Hybrid Edition to help enterprises optimise and modernise applications for hybrid cloud.

“It helps you to move monolithic applications into microservices, and it uses AI developed for IBM Research to analyse large application portfolios and provide recommendations for how to do it,” Whitehurst said.

He added that the technology is based on IBM’s Project Debater set of natural language processing technologies that understands meaning and applies the required source codes in application modernisation efforts.

The most notable AI development from IBM announced at Think 2021 is arguably Project CodeNet, a large dataset aimed at teaching AI how to code.

It comprises some 14 million code samples and about 500 million lines of code in more than 55 different programming languages including Java, Python, Go, as well as legacy languages like Cobol.

By using AI to translate between programming languages, Project CodeNet is touted to speed up the code migration process when refactoring monolithic applications. “Mechanistic ways to do translations can only get you so far – AI is making those translations much easier,” Whitehurst said.

Whitehurst said IBM’s latest efforts in AI are aimed at addressing the barriers to AI implementations. Citing a new IBM study, he noted that while AI adoption was flat last year, there has been a major uptick in AI investments now as organisations move into the post-pandemic world.

According to the study, 43% of respondents have accelerated AI initiatives, but 39% are still grappling with limited expertise while 32% are facing increasing data complexity caused by data silos.

Nick McQuire, chief of enterprise research at CCS Insight, said Watson Orchestrate and initiatives like Project Debater will make Watson more appealing in reaching the last mile of users that have so far evaded IBM’s AI.

“The path ahead in this area won’t be easy,” McQuire said. “Particularly, when you consider the user platforms that Microsoft and Google Cloud enjoy with their SaaS [software-as-a-service] applications – now a huge breeding ground for their AI by comparison – as well as the strategy that AWS is taking with its AI services.”