IBM Think 2023: CEO Krishna On The Need To Be ‘Careful’ With AI As An ‘Explosion Of Innovation’ To Come In Next Decade

AI is still not at the point of autonomous systems with no humans watching and taking control, says IBM CEO Arvind Krishna, as made-up information from AI presented as fact is still an issue. ‘You have to be careful. If the consequence is massive, I would be a little cautious.’

IBM Chairman and CEO Arvind Krishna was front and center during the vendor’s Think 2023 conference, using multiple speaking appearances to weigh in on IBM’s opportunity in artificial intelligence, hybrid cloud and quantum computing. He also spoke about Google’s role in AI, complimented VMware and forecast where AI is headed next.

The CEO of Armonk, N.Y.-based IBM also said the company had taken an incorrect approach to selling AI after achieving worldwide attention for a computer system that could beat world champion chess player Garry Kasparov in the 1990s and could beat champions of the game show “Jeopardy” in the 2010s.

The vendor assumed that customers would want AI black boxes where they didn’t see how deep learning systems could make decisions.

“That was an incorrect conclusion from the right observation,” Krishna said. “That was wrong.”

IBM CEO Krishna At Think 2023

He also warned against trusting fully autonomous AI systems with no human oversight and control in certain areas.

“You have to be careful,” he said. “If the consequence is massive, I would be a little cautious.”

Think 2023, a gathering of about 4,000 business and technology leaders, runs through Thursday in Orlando, Fla.

For the first time, IBM held a Partner Plus Day for content geared toward about 750 partners in attendance. Krishna also spoke during that event Monday.

IBM has about 55,000 worldwide channel partners, 12,000 of them in North America, according to CRN’s 2023 Channel Chiefs list.

Here’s what else Krishna had to say.

IBM’s AI Mistake

Krishna told the crowd of partners that although IBM was once at the forefront of AI with its chess world champion-beating system Deep Blue and its “Jeopardy” champions-beating question-answering computer system IBM Watson, the company wasted years on black box AI offerings.

The vendor did not anticipate that customers would want the ability to see inputs and operations.

“That was an incorrect conclusion from the right observation,” he said. “That was wrong.”

Krishna said that the use of a foundation model to deploy AI at scale is “a massive advantage.” But AI is still not at the point of autonomous systems with no humans watching and taking control. “Hallucinations” and made-up information from AI presented as fact is still an issue.

“You have to be careful,” he said. “If the consequence is massive, I would be a little cautious.”

More ‘Intentional’ With Hybrid Cloud

A change Krishna wants to see with IBM and partners is making customer adoption of hybrid cloud more “intentional” than incidental.

About 77 percent of enterprises surveyed by IBM said they were hybrid after the fact. If hybrid cloud is a popular destination, IBM and partners should get more prescriptive about it. VMware has been successful in marketing its offerings as ways for customers to use public cloud and on-premises, he told the crowd during Partner Plus Day.

“We need to make it intentional as opposed to they arrived there,” Krishna said.

IBM AI Will Stay B2B

Krishna told the crowd of analysts and reporters Tuesday that he views AI as in the first inning of its potential, with cloud in its fifth or sixth and quantum is “not even in the minor leagues, but it has the potential.”

He told the crowd that IBM will stay business-to-business (B2B) with its AI offerings, not offering consumer-focused AI such as the ChatGPT text-generator tool that ushered in hype around AI late last year. When it comes to B2C, “we are not necessarily very good at that,” Krishna said.

He also sees consulting services around AI as a revenue driver for IBM.

IBM will also meet customers with the pricing model they want, whether that is paying based on a software stack or for inferences and usage. “I think both are going to be there,” he said.

IBM will not make a Red Hat-level acquisition to boost its AI capabilities, Krishna said. The $34 billion Red Hat acquisition closed in 2019 and was a “once-in-a-decade, maybe once-in-two-decades” purchase, he said. But IBM will pursue other acquisitions with a focus around software, automation, hybrid cloud, cybersecurity, AI and machine learning, and consulting.

For consulting, acquisitions to boost its SAP, Salesforce, Amazon Web Services, Oracle and Adobe capabilities are of interest, he said.

When asked about IBM exploring AI for migrations, Krishna said the technology is in the works but “I wouldn’t call them ready for prime time.”

Eventually, IBM will offer ways for customers to convert COBOL and other legacy coding languages into modern languages at a massive scale.

“Will there be AI tools by this time next year that will help in migration of legacy code to modern code? Absolutely,” he said. “Will there be tools that will help you maybe change an encryption algorithm of AES or—heaven forbid—DES? … Absolutely. Will we have tools that can scan all of your code and recommend these are the parts that do need modernization? Absolutely. And we do work on all those.”

When asked about how to talk to customers who are interested in digital transformation but are concerned about the costs amid an economic slowdown and high inflation, Krishna said he usually suggests to customers that they find “a champion” inside the company who then tries to automate a dozen or so processes and measures the outcomes within a year.

During Krishna’s keynote address Tuesday, he said that from 2017 to early 2023, the rate of AI adoption by businesses doubled to about 30 percent or 40 percent and estimates $16 trillion in productivity will come from AI. “AI is going to unlock as much for your businesses as” the internet and ecommerce and cloud and mobile technologies.

For IBM, the opportunity in AI for businesses is in changing certain tasks in human resources, supply chain, inventory and the process of configure, price, quote (CPQ). Businesses will need accurate answers, the ability to deploy hundreds and thousands of models, avoid drift and trust the data.

“That is what we want to do,” he said. “It will also, by the way, get rid of a lot of errors that otherwise creep in because of all the issues around—whether you want to call it ‘fat fingers,’ or errors or people who are tired—all those things. That’s one big, big element.”

He estimated that 70 percent of people work on operations and maintenance. AI promises to cut the number down to 30 percent, allowing those people to work on innovating instead. Eventually, AI will remediate security issues.

AI is also necessary to fill the employment gap in cybersecurity and to handle the high volume of attacks and data to sift through in security. “If you don’t use artificial intelligence, there is no chance that your team of cyber analysts can scale to go address those issues,” he said.

AI that can show reasoning as opposed to memorization and pattern recognition is about five to seven years out, he said. “We’re going to see an explosion in innovation in the next decade, way more than we actually have seen in the previous couple.”

AI Job Loss Clarification

When asked about a recent report on IBM exploring AI and automation to replace thousands of jobs, Krishna clarified that he also expects new jobs to arise, with IBM growing in the number of AI engineers, data scientists, client engineers and other roles.

Thirty percent of back-office jobs will be obsolete in five years, he said. And in other fields, even when robots replace a task, a human still needs to oversee the robots.

“Manufacturing jobs are supposed to be evaporating—that’s not true,” he said.

Generative AI adoption is also accelerating the need for multi-cloud environments, which helps IBM, he said. Businesses with sensitive data and seeking privacy will still avoid public clouds. And some regulatory bodies are demanding multiple clouds or a mix of cloud and on-premises for resiliency, Krishna said.

He also discussed the importance of diversity and equity that needs to happen in physics, math, engineering, computer science and other science fields in education and hiring.

“I’m aghast that even in a country as advanced as the United States, the number of women in engineering is now at 20 percent or less,” he said. “I give a lot of credit to the life sciences. Biology has managed to get up to about 50 percent.”

Krishna On AI Competition

When asked about a recently leaked Google memo about Google’s need to embrace open-source AI technology and the lack of a moat around AI for Google or Microsoft-backed ChatGPT creator OpenAI, Krishna said that the memo is “dramatically underplaying” Google’s moats around advertising and a search engine that captures information up to the minute.

The possibility that users migrate from a search bar to a generative AI interface as the main method of using the internet is “the danger to” Google, he said. “That’s not a moat in a business. That’s about is the underlying fundamental model changing here?”

For IBM, Krishna described its moat as a trusted business for working with customer data, a developer of quantum computing and going all in on AI foundation models as some moats.

Hybrid Cloud Debate ‘Over’

During Krishna’s keynote address on Tuesday at Think, the IBM CEO said that 77 percent of businesses prefer hybrid cloud architecture for the ability to work with multiple public clouds, ensure data privacy, add in private cloud, work at the edge and still use on-premises resources.

“It has gone over the last four years from being a choice to be the predominant architecture choice for our clients. … That debate about that being the destination, I will assert, is over,” he said.

“We assert that a hybrid cloud choice gives you two and a half times more value than picking a singular answer from one of those underlying landscape choices.”

Krishna On Quantum

Although still in its early days, Krishna told the crowd during his keynote address that quantum isn’t science fiction. He already sees use cases in electric vehicle batteries, carbon sequestration, traffic routing, materials, search and finance.

“IBM has over 20 actual quantum computers,” he said. “Not simulators, not in software, not faked with something else. Real quantum computers that run on our cloud and are accessible to anybody who wants to go play with them.”

Still, quantum’s ChatGPT moment is still three to 10 years away, he said. He urged companies to take quantum cracking of encryption seriously.

He urged partners to take quantum seriously during Partner Plus Day as well.

“If quantum computing was never to exist, today’s encryption schemes are perfect,” he said. “We’ve got to get ready for the post-quantum era.”

Krishna On The Environment

During Partner Plus Day, Krishna implored partners to look into energy savings, if not for the societal benefit of helping the environment, then for the ability to run businesses better as energy costs grow.

“In this case, profit and purpose go together,. My personal view [is] 30 percent of energy use is completely and utterly wasted,” he said.

“If you aren’t paying attention to the environment, it’s OK. I guess you will make less profit.”