Predictions 2025: There is no easy button for AI — Prepare for grind!

AI made a side entry into the enterprise this year and spread rapidly, exceeding our 2024 AI prediction that 60% of workers will bring their own AI (BYOAI) in 2024. But, BYOAI is not a strategy. Enterprise AI leaders now realize that for longer-term AI success, they need an effective strategy combining data and AI to ensure AI enters through the front door. 

Forrester’s Q2 AI Pulse Survey, 2024 shows many concerns about generative AI (GenAI) use at US organizations, such as hallucinations, finding quality training, and challenges around governance and data protection. This accentuates the need for a strategy that evolves with the rapid pace of innovation. 

The strategy must include estimating business impact and ROI, selecting the right use cases, plans for cleaning and governing data, aligning on an operating model, training talent, experimenting with new application architectures, partnering internally and externally, and balancing risk and reward. This will not happen overnight. 

In the year ahead, you’ll need to put your nose to the grindstone to develop an effective AI strategy and implementation plan. In 2025, organizational success will depend on strong leadership, strategic refinement, and recalibration of enterprise data and AI initiatives commensurate with AI aspirations. We predict that:

* Most enterprises fixated on AI RoI will scale back prematurely.
Enterprises are achieving improved customer experience, employee productivity, and even new revenue streams with AI use. But an AI reset is underway. Obvious use cases that enterprises experimented with last year are now table stakes and embedded in business software. Leaders are realizing that ROI from investments will take longer than they anticipated, and are shifting toward pragmatically delivering RoI over time. 

In Forrester’s Q2 AI Pulse Survey, 2024, 49% of US GenAI decision-makers said their organization expects ROI on AI investments within one to three years and 44% said within three to five years. Impatience with AI RoI could prompt enterprises to prematurely scale back investments, which would be a long-term disadvantage. 

Instead, AI leaders must establish a solid strategy aligned to their business model and aspirations. Pick differentiating uses cases leveraging company-specific data and expertise, and create a roadmap that balances short-term and longer-term business ROI to create a flywheel for reinvesting early successes into future AI projects.

* 30% more large-firm CIOs will bring CDOs into the IT fold. 
CIOs will be tapped by CEOs to lead AI technology efforts and will realize the need to bridge the gap between technical and business expertise. Enter the chief data officer or (CDO). CDOs and other senior data and analytics leaders are chameleons, equally comfortable in an IT organization or in a line of business. 

Forrester’s State Of Data, Analytics, Measurement, And Insights Survey, 2024 shows that 39% of respondents note the most senior data and analytics leader reports to the CIO, while a similar share (37%) say they report to the CEO. The percentage of data leaders reporting to CIOs has increased throughout 2024. 

As CIOs lead the AI strategy, they will recognize that AI is often fraught with challenges that stem from problems with data foundations and stakeholder collaboration. To meet those challenges in 2025, CIOs will not only seek out CDOs as partners to assure AI success, but also integrate them into the IT fold, ensuring AI is a business strategy, not just a technology strategy. 

To be successful in AI, CIOs and CEOs must elevate the CDOs to a prominent position — not just as liaisons, but as true AI leaders who address change management, AI outcomes, and risks and who deliver strong RoI.

* 40% of highly-regulated enterprises will combine data and AI governance. 
The enforcement of the EU AI Act in February 2025, with fines of up to 7% of firms’ global turnover, and a potential AI tax and fines in the US will require integrated governance. Since AI is a data app and AI governance is data governance, 40% percent of highly-regulated companies will combine their data and AI governance programs. 

Compliance capabilities depend on visibility into how AI models use data and the nature of the data itself. With no universal templates, standards, or certifications, companies must urgently engage with third parties to create a new shared responsibility model and avoid repeating past data governance failures, as AI governance is highly complicated by its rapid pace and complexity. Governance professionals will need to broaden their skill set to cover both data and AI. 

Further, investments will flow into technologies that offer combined data and AI cataloging, access control, policy management, quality control, and risk management. To succeed, enterprises must build a cohesive governance framework, adopt or adapt governance solutions, promote collaboration, and automate compliance processes.

* AI pendulum will swing back to predictive AI for over 50% of use cases. 
According to Forrester’s Q2 AI Pulse Survey, 2024, the distribution of enterprise use cases leveraging predictive AI (36%) is roughly equal to those leveraging GenAI (35%). However, as enterprises hit transient roadblocks in applying GenAI in ways that meet expectations, organizations will double down on predictive AI applications. 

Tried and true predictive use cases like predictive maintenance, customer personalization, supply chain optimization, and demand forecasting will siphon investment dollars away from genAI-specific use cases in 2025. Forward-looking enterprises, however, will realize that predictive and generative AI are not mutually exclusive. Predictions can make the output of genAI even smarter, leading to an increase in the proportion of use cases leveraging both technologies from today’s 28% to 35%.

* Three out of four firms that build aspirational agentic architectures on their own will fail.
AI agentic architectures were a top emerging technology for 2024, but they’re not ready yet — expect another two years before they have any chance of meeting inflated automation hopes. Meanwhile, agentic AI is all the rage as companies push genAI beyond basic tasks into more complex actions. 

The challenge is that these architectures are convoluted, requiring multiple models, advanced RAG stacks, advanced data architectures, and specialized expertise. Aligning these models for focused outcomes is an unresolved issue that will disappoint eager developers. 

As a result, 75% of enterprises that attempt to build these agents themselves next year will fail and will turn to consultancies to build custom agent setups or will use agents embedded in their existing vendor software ecosystems. Savvy firms will grasp current limitations and lean on their vendor and systems integrator partners to build agents at the cutting edge of this technology.

Summary
The AI frenzy played out in the hallways of both vendors and enterprises in 2024. Rightfully, concerns around business outcomes came up as costs of experimentation spiraled and real-life benefits were slow to materialize. AI leaders realized they needed to prioritize business outcomes, clean their data houses, and start building AI talent. 

Forrester predicts 2025 will bring a renewed focus on strategy, deepened partnerships between business and IT, a pivot back to predictive AI, and new technologies and architectures as enterprises govern data and AI together.

-- Forrester Research, USA & Australia.