How should you think about AI as an IT leader in 2025

Despite how prominent the rise of AI has been, CIOs in particular have been understandably cautious about experimentation. According to a survey by BMC, about 94% of the respondents have said that AI has become a part of their IT strategy but only 17% have moved beyond the experimental phase, and only 5% have reportedly moved beyond the mature implementation phase to seeing tangible results.

Other C-suite leaders are looking at CIOs as data experts to recenter the organizational compass with AI initiatives while CIOs have their plate overflowing with budget constraints, evolving leadership landscapes, incapable legacy systems, and a shortage of skills. Employees are exploring AI in IT without much guidance while leaders aren’t really sure of how to communicate realistic expectations to upper management.

IT infrastructures can be rigid and often difficult to transform, which adds on to existing difficulties that pose a steep enough climb already. According to another survey by PwC, about 58% of the CIOs who participated are investing heavily in AI but about 82% of them say that achieving value from adopting these new technologies is a challenge in itself.

There’s a lot that CIOs still don’t realize about AI but an urgent need to “adopt or be left behind” raises another dilemma – if not now, then when? The struggle to derive significant value from AI stems not just from its unpredictability but also from various ineptitudes that IT teams face, often due to flawed innovation strategies. In the next couple of years (as 2026 ends), over 90% of organizations are predicted to experience an IT skills crisis with not enough talent to fill essential positions. This threatens an already overworked department with further workload, begging the question – should AI be that much of a priority for CIOs and IT leaders?

Anticipate a change in your leadership style

CIOs are under a lot of pressure to change the way they work and lead. But a change in leadership approach doesn’t only demand a change in strategy, it demands a change in mindset. A lot of leaders in IT today have become technology veterans with over a couple of decades worth of experience under their belt. While such leaders have seen what can be considered as the golden era of technology, they can be a bit resistant to change.

The reason, however, is arguably justifiable – senior executives who’ve undertaken a few IT transformation initiatives have seen at least one major transformation that failed. Employees who were part of such initiatives have expressed negative emotions almost significantly more since failed attempts can mean layoffs or stalled appraisals. It’s true that not all companies can afford such failures but there’s no way around evolution without it.

The pace at which digitization has advanced in less than 5 years has almost been unbelievable. The fact is, ignoring the need for a change in leadership style despite everything that’s stacked against it will definitely be worse than the alternative. So, then what’s the answer? Traditional leadership can be limiting. Your employees need more autonomy; more freedom to implement fresh ideas. The ideas can be questioned but not reprimanded. Work will increasingly need to become more collaborative and be spread into networks rather than moving through strict hierarchies. Now more than ever, your responsibility as a leader is to enable connections and move out of the way.

This may sound straightforward but with a technology that’s as nascent as AI, leaders want to be on top of everything. That’s not entirely wrong but in the process, this approach might lead to silos, micromanagement, and the inability to see the big picture stuff.

AI adoption is a marathon, not a sprint

The most important thing at the moment is to set realistic goals and expectations from AI projects in IT. A general misconception seems to be floating around about AI being able to solve problems of all magnitudes, especially among non-technical leaders who don’t understand certain intricacies. At best, AI can be described as more of a purpose-driven technology that eliminates specific redundancies leading to productivity gains. But despite being a self-learning model, AI is far from perfect. If anything, AI needs more human surveillance and oversight now more than ever. Controlled implementation and quality assurance will lay the groundwork for whatever comes next.

As a leader, you must think of AI as an augmentation tool that enhances human capabilities. Some roles and responsibilities will become vestigial as AI proliferates and that’s unavoidable, but in the long run, organizations will be able to scale beyond existing benchmarks. The truth is, agentic AI, AGI, and Gen AI have had a bittersweet effect so far with some people favoring it while others not so much. There’s no right or wrong here, just perspectives. In IT, where employees are already knee-deep in resolving current concerns, abruptly introducing a new technology like AI might be more devastating than beneficial if we consider the following scenarios:

• Employees face more workload with changes in infrastructure, processes, and tech stack; the workforce isn’t happy about the change and resent management; transformation fails, and all invested resources don’t translate into improved ROI for the business.
• Employees face more workload with changes in infrastructure, processes, and tech stack; the workforce isn’t happy about the change and resent management; transformation, however, is successful, and the business recognizes more ROI from said changes.
• Employees welcome change and are happy about it; not a lot of infrastructural changes are required and eventually, there’s less workload on the workforce; transformation, however, isn’t effective, and business objectives don’t align with expected ROI.
• Employees welcome change and are happy about it; not a lot of infrastructural changes are required and eventually, there’s less workload on the workforce; transformation is successful, and the business recognizes more ROI from said changes.

3 of the 4 scenarios above (75%) result in a disruption that’s either not beneficial for the team, the organization, or both. This is why an abrupt change to accommodate AI in IT almost certainly won’t work for you. Hence, taking things slow is the best bet.

• First, test the waters, have discussions with the C-suite and with your team to understand how they think of AI and how you can bridge the gap between what’s expected and what’s possible.
• Don’t force change but rather spend time understanding how you can enable your team to embrace it.
• Communicate realistic outcomes from these changes so everyone in the organization is aligned on the specifics.
• Experiment on a smaller scale, implement when it feels ready, and scale things that work.

AI in IT won’t be much without data. A lot of CIOs are well aware of this – about 47% of them stress that data transformation is their focus area, placing it above customer relationship management, support, and marketing. About 32% also believe that data management is the top challenge for CIOs and IT teams tied with handling legacy systems. A lot of organizations have more data than they can make sense of, and this often leads to data silos, unorganized structures, and untapped potential. With AI, it’ll be easier to train LLMs on unexplored data and hit two birds with one stone (data transformation and AI initiatives). There’s also a question of trustworthy and secure AI and not all organizations will be comfortable sharing sensitive data as training models. This could mean selective and regulated AI implementation in-house.

Cybersecurity will also continue to be equally important. The past decade has seen some of the worst ransomware attacks and security breaches, strengthening the need for better security protocols. The unpredictability of these breaches weighs heavy on IT leaders and the best they can do is invest in better tools and train company staff in safety measures. Preempting a security breach is perhaps the smartest way to address this concern. Half-measures to accommodate AI will leave cracks in your IT network, making it susceptible to attacks – you need to avoid this as strictly as you can.

Stacked-up technical debt continues to remain as yet another hurdle for change. Organizations still allocate upwards of 70% of their IT budget to handling technical debt, making it harder for the team to think of innovation. While legacy systems remain that drain employee resources, any room for transformation is compromised.

The current state of AI in IT

A trillion-dollar solution for a problem that doesn’t exist – Goldman Sachs has explored what leaders think about AI and a common consensus seems to be against that of AI. AI as a technology isn’t designed to solve the complex problems that would justify the costs and what we’re seeing right now is a myriad of resources being thrown for the development of AI. Another study, in partnership with the Upwork institute of research, found that about 77% of employees view AI as an extension to their workload instead of a technology that eases it. What’s worse, most employees don’t know how to meet the productivity expectations of their employers while they struggle to keep up with new tools and workflows.

The state of AI adoption across industries as reported by Goldman Sachs

These views, however, are contradictory as a lot of companies have reported productivity gains as well. Honeywell – a conglomerate that primarily deal in aerospace, industrial, and energy automation – has spent the better part of the last few years developing its data-driven enterprise and is now ready to leverage a layer of AI atop this data goldmine across every department of its business.

The Fortune 500 giant has implemented AI copilots to automate IT help desk requests, reducing manual ticket resolutions by 80%. An internal chatbot has been trained on hundreds and thousands of manuals and internal articles to answer questions in real-time, saving hours of work for their employees. A workforce as strong as 95,000 now has access to more than a dozen AI use cases already in production. They moved way beyond the experimentation phase and have implemented an organization-wide AI initiative with IT being at the core of it.

The data and IT leaders at Honeywell expressed that the reason why they were able to get AI-ready and scale things this quickly is because they figured out their data strategy first; there is no AI strategy without a data strategy. They’re focused immensely on their data enterprise warehouse (EDW) and with that in place, it’s easier to source the training materials necessary for dependable and clean AI.

There’s also a strong argument for the development of cleaner, safer AI for which the supporting infrastructure needs to be built in-house. This does require strong engineering overhead for 6-12 months along with cost-intensive initiatives, but the payoff is worth it. It might be a tight rope to walk but companies like ServiceNow have been able to develop scaleable AI solutions not just for IT but for functions across the board. Smaller companies that cannot afford such initiatives are relying on smaller AI startups that have emerged uncontrollably in the past couple of years. Moderne.ai - a product that eliminates technical debt – is an example of such tools that couldn’t have existed before the AI boom. Everyone else has adopted one of the major players in the AI space – Microsoft AI copilot, Google, or OpenAI – in some form or another.

AI in IT seems to be a transitional state across a spectrum of organizations from those who’re seeing no ROI to those who’re already seeing tremendous impact on revenue. According to a correlations report, larger enterprise organizations (2001 – 5000 employees) are more likely to see negative ROI from AI investments as compared to SMBs. But when it comes to IT budget allocation, smaller organizations are far more skeptical of their AI spends since they are least likely to spend more than 20% of their IT budget on AI while larger enterprises are most likely to (and have already) allocated more than 20% to AI.

AI spend and ROI also factors in the location – apparently, European companies are lagging far behind their US counterparts by 45-70% in terms of AI adoption across various sectors of different sizes. For sectors that are significantly larger in the US than Western Europe (software, pharma, healthcare, etc.), there seems to be an even more pronounced disparity upwards of 70%.

Disparity between external AI and IT spend in Europe vs. US

Disparity between internal AI and IT spend in Europe vs. US

What are the most probable reasons for this?

• Shortage of skill: A general lack of skilled professionals has created a pressure system that requires upskilling and reskilling efforts of about 12 million employees to improve productivity through AI-led projects.
• Compensation and career growth: Europe has more AI professionals than the US but of the 22% who study in Europe, only 14% work there because US companies offer better pay and growth.
• Why is US offering better compensation than Europe? Because of significantly more VCs and incubators that are ramping up AI investments in the country.
• Complex regulations: The EU AI Act isn’t easily comprehensible — about 70% of the companies find them to be too complex.

European companies also seem to be investing less willingly in their IT teams. According to another report, larger firms in Europe spend 10% less on IT than those in the US and, what’s more, not enough IT leaders don’t see this as a concern. With stringent budget allocations to IT in general, it isn’t a surprise that a much smaller piece of the pie is shared for digital transformation initiatives such as AI.

Smaller to medium-sized companies have slightly lost trust in AI whereas larger enterprises with over 5000 employees have reportedly started to trust AI more over the past year. Interestingly, trust was higher in companies where AI projects were initiated by IT teams as compared to those where AI projects were more of a C-suite undertaking. This could be because IT leaders have more of a responsibility to deliver business outcomes like ROI than anyone else on the team. It can be inferred from these statistics that larger organizations have more appetite for risk than smaller ones and rightfully so. This just goes to show that AI needs more space and patience for growth. Instant returns are a daydream for any new initiative and nothing substantial prevails without associated risks.

This begs the question – what do you think about AI in IT? Do you think that:

• AI is still in its nascent stages and there’s not much value in it anyway. Europe isn’t missing out at all.
• Companies have reportedly seen more workload with AI than productivity—IT teams have to learn new technology on top of an existing tech stack.
• AI isn’t the messiah everyone thinks it is—most use cases are surface-level and don’t translate to any tangible productivity or ROI. T1 support will likely get completely automated at best but nothing more.
• AI is still a ‘good to have’ and not a 'must have'. Most probably, it will continue being that way for a while.
• Tools like Moderne.ai — a product that helps eliminate tech debt — are simply extensions of a problem that can be solved without AI.
• AI is a bubble that will eventually pop, only the top corporates will survive, and most AI startups will go under or squander more resources in an attempt to go back to their pre-AI roots.
• IT budget should be spent on more important problems like addressing legacy systems, upskilling employees, strengthening IT infrastructure, etc.

Or do you think that:

• AI is a revolutionary technology that will completely change the way IT teams work with data.
• Yes, AI is in the nascent stages but tools that exist now couldn’t have been possible without it.
• About 70% of IT teams have already implemented AI chatbots and helpdesk. AI use cases are prolific and will continue to become more effective.
• Developing cleaner AI in-house like ServiceNow is the safest bet. AI that develops as a result will not only be clean but dependable.
• Important aspects of IT like cybersecurity, infrastructure, and analytics will become safer, more agile, and more robust with AI.
• With the right upskilling and reskilling, IT teams should become more productive than ever with AI.
• Europe is missing out, but restrictive AI laws might push for development of cleaner AI.

What might IT look like in a few years?

Most likely, entry-level IT support will be replaced entirely by automation and AI chatbots for internal troubleshooting (about 70% of IT leaders have already started using Gen AI for chatbots and AI helpdesks). Employees might even be able to troubleshoot basic operations themselves such as onboarding, remote actions, etc. Organizations have reportedly already implemented helpdesk AI chatbots for internal IT support that can pull previous answers, respond to user queries directly, resolve tickets without manual intervention, and handle system alerts. What’s more, T1 support functions might be completely automated, resulting in cost savings for IT leaders. Although controversial, this might arguably be a realistic scenario that aligns closely with IT transformation initiatives and even business objectives.

Positions like cybersecurity professionals, solution architects, and implementation analysts will continue to be in demand but might also witness a lot of outsourcing or even staff augmentation. We’ll see more AI bots reporting to cybersecurity experts while IT leaders realize how expensive it is to maintain full-time SOCs. Unfortunately, flourishing AI might even affect high-paying jobs, creating a domino effect that makes the vendor market more competitive i.e. more outsourcing, more innovative vendors, competitive pricing, etc.

Whatever said and done, IT will continue to be a department that requires a hands-on approach and any new technology that disrupts ongoing processes will need constant monitoring. While there’s certainly a possibility of some jobs becoming obsolete, there’s also an undeniable possibility of new (and in-demand) careers being born from these paradigm shifts some of which have already surfaced like prompt engineers and AI & ML engineers.

Brace yourself; pace yourself

Things have never been this unprecedented. Everyone seems to be more intrigued than excited about what’s next. AI chatbots have become synonymous with support while they were almost unheard of less than a couple of years ago. It’s remarkable, to say the least, how adapting we’ve been to these changes – some have been more critical about this than others, while some simply want to adapt quickly from fear of being left behind. You don’t want to be moving too fast or too slow, but you need to figure out your pace with time.

Success ultimately depends on the people responsible — you, your team, and your ability to communicate critical factors to other decision-makers. Break down data silos, improve transparency, and empower your team to voice concerns. The way you define your IT budget directly impacts your ability to support transformation initiatives. A healthy approach involves transparent discussions and thoughtful budget allocation. As far as budgeting is concerned, we have a detailed template to assist you with a solid structure.

Use our template to set your IT budget for 2025

IT leaders are torn between AI being a revolution and AI being an exaggerated new toy for companies to play with. There is, however, a leaning towards AI being more than just a bubble waiting to pop and the whole work forgetting about it. Some things will irrevocably change because of it and resisting will only make it worse. More than anything, it demands a mindset change over a perspective change. Thinking of AI as an augmentation tool is easier than thinking of it as a replacement tool. Yes, AI isn’t solving the big problems right now, but it might eventually will. As an IT leader, you need to upskill and prepare your team for gradual changes. AI is not worth going head over heels crazy for but it’s not something you should ignore either. Its ubiquitousness definitely calls for having an open mind at least.

‍