CIOs tackle talent strategies, org structures as AI takes hold
This story was originally published on CIO Dive. To receive daily news and insights, subscribe to our free daily CIO Dive newsletter.
Cambridge, Mass. — The rapid pace of AI development is putting CIOs in a tricky spot: Not only are they responsible for deploying tools that can boost productivity, they're also working to ensure the tools can be utilized once deployed. The dynamic has made workforce upskilling efforts a priority.
It's up to tech leaders to help shape a culture that enables AI experimentation, according to Monica Caldas, global CIO at Liberty Mutual Insurance.
"I do not believe that AI thrives in heavily authoritarian, top-down environments," said Caldas, speaking last week at the MIT Sloan CIO Symposium. "I think the way people pick it up is through play."
Along with encouraging experimentation, the risk profile of AI necessitates guardrails as businesses tackle change management.
"It's not anarchy, but it's also not authoritarian," Caldas said. "You have to hit that sweet spot, and that's where adoption really starts."
Other businesses are preparing their employees through targeted training programs. More than half of leaders said they plan to upskill their workforce ahead of AI implementation plans, according to a January survey from Revature. More than 4 in 5 decision-makers flagged access to talent as a top concern.
Potential productivity wins can help existing staff embrace upskilling efforts, said Dimitris Bountolos, chief information and innovation officer at infrastructure company Ferrovial.
"What we have seen is an excitement of staff to be self-sufficient in activities that were really bureaucratic," Bountolos said.
The deployment of generative AI tools caused a rush of interest in prompt engineering roles. Interest has since fizzled, as businesses began to understand that learning to prompt generative AI systems is a core skill that should be developed more broadly.
AI savviness should be embraced by the entire organization, according to Reshmi Ramachandran, head of partnerships and go to market strategy at consulting firm Cprime.
"When we consult with companies we often tell them: never do prompt engineering in isolation. It's not an isolated job, it is actually a cross-functional skill," Ramachandran said. "You get some of your best prompts from marketing leaders, from HR, because that's where the context is."
In addition to changes in job functions, departmental structures are also evolving.
The wave of AI adoption is helping to accelerate a shift away from the established pyramid-shaped organizational structures in software development, according to Aamer Baig, senior partner, Chicago, at McKinsey & Company.
"In the last decade or so, we've proven that is not the most effective and economical way of delivering software," said Baig. A diamond-shaped model with a team of somewhere between eight to 12 was identified as the most effective.
But with the influx of agentic AI, that organizational structure is also changing.
"Now, we have a new model, which is enabled and powered by AI, that has a product person, product builders and many, many agents to support, which can deliver as much output as a diamond-shaped team does," Baig said.
In addition to serving as CIO, tech executives will need to take on additional roles including "chief influencing officer, chief change management officer" as organizations adjust to shifts in their core talent and operational structures.
"The ability to move that sort of organization and that complexity forward will differentiate the winners and the losers in large companies," Baig said.
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Associated Press
34 minutes ago
- Associated Press
States are rolling out red carpets for data centers. But some lawmakers are pushing back
HARRISBURG, Pa. (AP) — The explosive growth of the data centers needed to power America's fast-rising demand for artificial intelligence and cloud computing platforms has spurred states to dangle incentives in hopes of landing an economic bonanza, but it's also eliciting pushback from lawmakers and communities. Activity in state legislatures — and competition for data centers — has been brisk in recent months, amid an intensifying buildout of the energy-hungry data centers and a search for new sites that was ignited by the late 2022 debut of OpenAI's ChatGPT. Many states are offering financial incentives worth tens of millions of dollars. In some cases, those incentives are winning approval, but only after a fight or efforts to require data centers to pay for their own electricity or meet energy efficiency standards. Some state lawmakers have contested the incentives in places where a heavy influx of massive data centers has caused friction with neighboring communities. In large part, the fights revolve around the things that tech companies and data center developers seem to most want: large tracts of land, tax breaks and huge volumes of electricity and water. And their needs are exploding in size: from dozens of megawatts to hundreds of megawatts and from dozens of acres up to hundreds of acres for large-scale data centers sometimes called a hyperscaler. While critics say data centers employ relatively few people and pack little long-term job-creation punch, their advocates say they require a huge number of construction jobs to build, spend enormous sums on goods and local vendors and generate strong tax revenues for local governments. In Pennsylvania, lawmakers are writing legislation to fast-track permitting for data centers. The state is viewed as an up-and-coming data center destination, but there is also a sense that Pennsylvania is missing out on billions of dollars in investment that's landing in other states. 'Pennsylvania has companies that are interested, we have a labor force that is capable and we have a lot of water and natural gas,' said state Rep. Eric Nelson. 'That's the winning combination. We just have a bureaucratic process that won't open its doors.' It's been a big year for data centers Kansas approved a new sales tax exemption on goods to build and equip data centers, while Kentucky and Arkansas expanded pre-existing exemptions so that more projects will qualify. Michigan approved one that carries some protections, including requirements to use municipal utility water and clean energy, meet energy-efficiency measures and ensure that it pays for its own electricity. Such tax exemptions are now so widespread — about three dozen states have some version of it — that it is viewed as a must-have for a state to compete. 'It's often a nonstarter if you don't have them, for at least the hyperscalers,' said Andy Cvengros, who helps lead the data center practice at commercial real estate giant JLL. 'It's just such a massive impact on the overall spend of the data center.' Zoning, energy fights often frustrate developers In West Virginia, lawmakers approved a bill to create 'microgrid' districts free from local zoning and electric rate regulations where data centers can procure power from standalone power plants. Gov. Patrick Morrisey, a Republican, called the bill his 'landmark policy proposal' for 2025 to put West Virginia 'in a class of its own to attract new data centers and information technology companies.' Utah and Oklahoma passed laws to make it easier for data center developers to procure their own power supply without going through the grid while Mississippi rolled out tens of millions of dollars in incentives last year to land a pair of Amazon data centers. In South Carolina, Gov. Henry McMaster signed legislation earlier this month that eased regulations to speed up power plant construction to meet demand from data centers, including a massive Facebook facility. The final bill was fought by some lawmakers who say they worried about data centers using disproportionate amounts of water, taking up large tracts of land and forcing regular ratepayers to finance the cost of new power plants. 'I do not like that we're making customers pay for two power plants when they only need one,' Senate Majority Leader Shane Massey told colleagues during floor debate. Still, state Sen. Russell Ott suggested that data centers should be viewed like any other electricity customer because they reflect a society that is 'addicted' to electricity and are 'filling that need and that desire of what we all want. And we're all guilty of it. We're all responsible for it.' Some lawmakers are hesitant In data center hotspots, some lawmakers are pushing back. Lawmakers in Oregon are advancing legislation to order utility regulators to ensure data centers pay the cost of power plants and power lines necessary to serve them. Georgia lawmakers are debating a similar bill. In Virginia, the most heavily developed data center zone in the U.S., Gov. Glenn Youngkin vetoed a bill that would have forced more disclosures from data center developers about their site's noise pollution and water use. In Texas, which endured a deadly winter blackout in 2021, lawmakers are wrestling with how to protect the state's electric grid from fast-growing data center demand. Lawmakers still want to attract data centers, but a bill that would speed up direct hookups between data centers and power plants has provisions that are drawing protests from business groups. Those provisions would give utility regulators new authority to approve those agreements and order big electric users such as data centers to switch to backup generators in a power emergency. Walt Baum, the CEO of Powering Texans, which represents competitive power plant owners, warned lawmakers that those provisions might be making data center developers hesitant to do business in Texas. 'You've seen a lot of new announcements in other states and over the last several months and not as much here in Texas,' Baum told House members during a May 7 committee hearing. 'I think everybody right now is in a waiting pattern and I worry that we could be losing to other states while that waiting pattern is happening.' ___ Follow Marc Levy on X at
Yahoo
34 minutes ago
- Yahoo
States are rolling out red carpets for data centers. But some lawmakers are pushing back
HARRISBURG, Pa. (AP) — The explosive growth of the data centers needed to power America's fast-rising demand for artificial intelligence and cloud computing platforms has spurred states to dangle incentives in hopes of landing an economic bonanza, but it's also eliciting pushback from lawmakers and communities. Activity in state legislatures — and competition for data centers — has been brisk in recent months, amid an intensifying buildout of the energy-hungry data centers and a search for new sites that was ignited by the late 2022 debut of OpenAI's ChatGPT. Many states are offering financial incentives worth tens of millions of dollars. In some cases, those incentives are winning approval, but only after a fight or efforts to require data centers to pay for their own electricity or meet energy efficiency standards. Some state lawmakers have contested the incentives in places where a heavy influx of massive data centers has caused friction with neighboring communities. In large part, the fights revolve around the things that tech companies and data center developers seem to most want: large tracts of land, tax breaks and huge volumes of electricity and water. And their needs are exploding in size: from dozens of megawatts to hundreds of megawatts and from dozens of acres up to hundreds of acres for large-scale data centers sometimes called a hyperscaler. While critics say data centers employ relatively few people and pack little long-term job-creation punch, their advocates say they require a huge number of construction jobs to build, spend enormous sums on goods and local vendors and generate strong tax revenues for local governments. In Pennsylvania, lawmakers are writing legislation to fast-track permitting for data centers. The state is viewed as an up-and-coming data center destination, but there is also a sense that Pennsylvania is missing out on billions of dollars in investment that's landing in other states. 'Pennsylvania has companies that are interested, we have a labor force that is capable and we have a lot of water and natural gas,' said state Rep. Eric Nelson. "That's the winning combination. We just have a bureaucratic process that won't open its doors.' It's been a big year for data centers Kansas approved a new sales tax exemption on goods to build and equip data centers, while Kentucky and Arkansas expanded pre-existing exemptions so that more projects will qualify. Michigan approved one that carries some protections, including requirements to use municipal utility water and clean energy, meet energy-efficiency measures and ensure that it pays for its own electricity. Such tax exemptions are now so widespread — about three dozen states have some version of it — that it is viewed as a must-have for a state to compete. 'It's often a nonstarter if you don't have them, for at least the hyperscalers,' said Andy Cvengros, who helps lead the data center practice at commercial real estate giant JLL. 'It's just such a massive impact on the overall spend of the data center.' Zoning, energy fights often frustrate developers In West Virginia, lawmakers approved a bill to create 'microgrid' districts free from local zoning and electric rate regulations where data centers can procure power from standalone power plants. Gov. Patrick Morrisey, a Republican, called the bill his 'landmark policy proposal' for 2025 to put West Virginia 'in a class of its own to attract new data centers and information technology companies.' Utah and Oklahoma passed laws to make it easier for data center developers to procure their own power supply without going through the grid while Mississippi rolled out tens of millions of dollars in incentives last year to land a pair of Amazon data centers. In South Carolina, Gov. Henry McMaster signed legislation earlier this month that eased regulations to speed up power plant construction to meet demand from data centers, including a massive Facebook facility. The final bill was fought by some lawmakers who say they worried about data centers using disproportionate amounts of water, taking up large tracts of land and forcing regular ratepayers to finance the cost of new power plants. 'I do not like that we're making customers pay for two power plants when they only need one,' Senate Majority Leader Shane Massey told colleagues during floor debate. Still, state Sen. Russell Ott suggested that data centers should be viewed like any other electricity customer because they reflect a society that is 'addicted' to electricity and are 'filling that need and that desire of what we all want. And we're all guilty of it. We're all responsible for it.' Some lawmakers are hesitant In data center hotspots, some lawmakers are pushing back. Lawmakers in Oregon are advancing legislation to order utility regulators to ensure data centers pay the cost of power plants and power lines necessary to serve them. Georgia lawmakers are debating a similar bill. In Virginia, the most heavily developed data center zone in the U.S., Gov. Glenn Youngkin vetoed a bill that would have forced more disclosures from data center developers about their site's noise pollution and water use. In Texas, which endured a deadly winter blackout in 2021, lawmakers are wrestling with how to protect the state's electric grid from fast-growing data center demand. Lawmakers still want to attract data centers, but a bill that would speed up direct hookups between data centers and power plants has provisions that are drawing protests from business groups. Those provisions would give utility regulators new authority to approve those agreements and order big electric users such as data centers to switch to backup generators in a power emergency. Walt Baum, the CEO of Powering Texans, which represents competitive power plant owners, warned lawmakers that those provisions might be making data center developers hesitant to do business in Texas. 'You've seen a lot of new announcements in other states and over the last several months and not as much here in Texas," Baum told House members during a May 7 committee hearing. "I think everybody right now is in a waiting pattern and I worry that we could be losing to other states while that waiting pattern is happening.' ___ Follow Marc Levy on X at Error in retrieving data Sign in to access your portfolio Error in retrieving data Error in retrieving data Error in retrieving data Error in retrieving data
Forbes
an hour ago
- Forbes
Is Flawed AI Distorting Executive Judgment? — What Leaders Must Do
The AI symbol sits at the heart of a circle formed by bright yellow foldable caution signs adorned ... More with exclamation marks. This image creatively conveys the urgent need for awareness and careful consideration of AI's rapid growth and its implications. The design's high impact, with its strong contrast and focal point, makes it an effective tool for raising awareness or sparking conversations around technology, security, and innovation. Perfect for customizable content with plenty of space for additional messaging or branding. As AI embeds deeper into leadership workflows, a subtle form of decision drift is taking hold. Not because the tools are flawed but because we stop questioning them. Their polish is seductive. Their speed, persuasive. But when language replaces thought, clarity no longer guarantees correctness. In July 2023, the Chicago Sun-Times published an AI-generated summer reading list. The summaries were articulate. The titles sounded plausible. But only five of the fifteen books were real. The rest? Entirely made up: fictional authors, fabricated plots, polished prose built on nothing. It sounded smart. It wasn't. That's the risk. Now imagine an executive team building its strategy on the same kind of output. It's not fiction anymore. It's a leadership risk. And it's happening already. Quietly. Perceptibly. In organizations where clarity once meant confidence and strategy was something you trusted. Not just in made-up book titles but in the growing gap between what sounds clear and what's actually correct. Large language models aren't fact checkers. They're pattern matchers. They generate language based on probability, not precision. What sounds coherent may not be correct. The result is a stream of outputs that look strategic but rest on shaky ground. This isn't a call to abandon AI. But it is a call to re-anchor how we use it. To ensure leaders stay accountable. To ensure AI stays a tool, not a crutch. I'm not saying AI shouldn't inform decisions. But it must be paired with human intuition, sense making and real dialogue. The more confident the language, the more likely it is to go unquestioned. Model collapse is no longer theoretical. It's already happening. It begins when models are trained on outputs from other models or worse, on their own recycled content. Over time, distortions multiply. Edge cases vanish. Rare insights decay. Feedback loops breed repetition. Sameness. False certainty. Businessman with white cloud instead of head on blue background. Businessman and management. ... More Business and commerce. Digital art. As The Register warned, general purpose AI may already be declining in quality, not in tone but in substance. What remains looks fluent. But it says less. That's just the mechanical part. The deeper concern is how this affects leaders. When models feed on synthetic data and leaders feed on those outputs, what you get isn't insight. It's reflection. Strategy becomes a mirror, not a map. And we're not just talking bias or hallucinations. As copyright restrictions tighten and human-created content slows, the pool of original data shrinks. What's left is synthetic material recycled over and over. More polish. Less spark. According to researchers at Epoch, high quality training data could be exhausted by 2026 to 2032. When that happens, models won't be learning from the best of what we know. They'll be learning from echoes. Developers are trying to slow this collapse. Many already are, by protecting non-AI data sources, refining synthetic inputs and strengthening governance. But the impending collapse signals something deeper. A reminder that the future of intelligence must remain blended — human machine, not machine alone. Intuitive, grounded and real. Psychologists like Kahneman and Tversky warned us long ago about the framing trap: the way a question is asked shapes the answer. A 20 percent chance of failure feels different than an 80 percent chance of success, even if it's the same data. AI makes this trap faster and more dangerous. Because now, the frame itself is machine generated. A biased prompt. A skewed training set. A hallucinated answer. And suddenly, a strategy is shaped by a version of reality that never existed. Ask AI to model a workforce reduction plan. If the prompt centers on financials, the reply may omit morale, long-term hiring costs or reputational damage. The numbers work. The human cost disappears. AI doesn't interrupt. It doesn't question. It reflects. If a leader seeks validation, AI will offer it. The tone will align. The logic will sound smooth. But real insight rarely feels that easy. That's the risk — not that AI is wrong, but that it's too easily accepted as right. When leaders stop questioning and teams stop challenging, AI becomes a mirror. It reinforces assumptions. It amplifies bias. It removes friction. That's how decision drift begins. Dialogue becomes output. Judgment becomes approval. Teams fall quiet. Cultures that once celebrated debate grow obedient. And something more vital begins to erode: intuition. The human instinct for context. The sense of timing. The inner voice that says something's off. It all gets buried beneath synthetic certainty. To stop flawed decisions from quietly passing through AI-assisted workflows, every leader should ask: AI-generated content is already shaping board decks, culture statements and draft policies. In fast-paced settings, it's tempting to treat that output as good enough. But when persuasive language gets mistaken for sound judgment, it doesn't stay in draft mode. It becomes action. Garbage in. Polished out. Then passed as policy. This isn't about intent. It's about erosion. Quiet erosion in systems that reward speed, efficiency and ease over thoughtfulness. And then there's the flattery trap. Ask AI to summarize a plan or validate a strategy, and it often echoes the assumptions behind the prompt. The result? A flawed idea wrapped in confidence. No tension. No resistance. Just affirmation. That's how good decisions fail — quietly, smoothly and without a single raised hand in the room. Leadership isn't about having all the answers. It's about staying close to what's real and creating space for others to do the same. The deeper risk of AI isn't just in false outputs. It's in the cultural drift that happens when human judgment fades. Questions stop. Dialogue thins. Dissent vanishes. Leaders must protect what AI can't replicate — the ability to sense what's missing. To hear what's not said. To pause before acting. To stay with complexity. AI can generate content. But it can't generate wisdom. The solution isn't less AI. It's better leadership. Leaders who use AI not as final word but as provocateur. As friction. As spark. In fact, human-generated content will only grow in value. Craft will matter more than code. What we'll need most is original thought, deep conversation and meaning making — not regurgitated text that sounds sharp but says nothing new. Because when it comes to decisions that shape people, culture and strategy, only human judgment connects the dots that data can't see. In the end, strategy isn't what you write. It's what you see. And to see clearly in the age of AI, you'll need more than a prompt. You'll need presence. You'll need discernment. Neither can be AI trained. Neither can be outsourced.
