AI scaling: top priority 2023-2025

According to MIT’s CIO vision 2025 report, the top priority for businesses over the next three years will be scaling up AI and machine learning. CIOs also plan to multiply the use cases in all core business functions.

Data and AI priority for companies. Nevertheless, observes the MIT Technology Review in its study ‘CIO vision 2025’, few still have an advanced level of mastery today. Only 1% of organizations can be considered “data driven”. They are determined to reach a milestone in maturity over the next three years.

For the decision-makers interviewed, the challenge over the 2022-2025 period will therefore consist in deploying use cases in all the major functions of the company. Thus, more than half expect the use of AI to be widespread or critical in IT, finance, product development, marketing and sales.

Read also: CDOs and DSI must help businesses to be data-centric

Wide, aggressive and enthusiastic use of AI

The AI ​​strategy of the Chief Digital Officer of Japanese insurer Tokio Marine is clear: It’s about “applying AI as broadly, aggressively and enthusiastically as possible. No part of our business should be spared from AI,” says Masashi Namatame.

In this perspective, the insurer is therefore striving to become “an AI-driven company” or “AI driven”. This increase in competence requires time and investment: “We are still learning from AI and trying to find more and more efficient ways to apply it to our business”, reacts the CDO.

In fact, the study points out that a significant increase in spending is planned to strengthen the Data and AI foundations of companies. And this also applies to the most advanced or “data leaders.” In these companies, spending on data security is expected to increase by 101% over three years.

Investments will also focus on data governance (+85%), new data and AI platforms (69%) and existing platforms (63%). This is more than for the average company with respective increases of 59%, 52%, 40% and 42%.

Thwarted ambition to scale up

Accelerating the use of artificial intelligence and use cases is not the only objective, however. The number one priority according to the study is therefore that of scaling up:

“More than three-quarters (78%) of the senior executives we surveyed – and almost all (96%) of the leadership group – say the scaling up of AI and machine learning to create business value is their top data strategy priority for the next three years.”

The study also observes a clear evolution in the nature of the business value generated thanks to AI. In 2022, 31% of respondents report an improvement in security and risk management, ahead of time-to-market (20%) and efficiency (14%).

MIT2 Review

In 2025, AI should result in increased revenue for 30% of decision makers, compared to 14% in 2022. 26% plan to improve efficiency and 16% to reduce costs. The direct financial impact of the use of AI is therefore clearly evident in corporate strategies.

To generate more value, they will therefore more than ever be confronted with the imperative of scaling up. But many organizations still face this obstacle, warns MIT.

“We spend a lot of time trying to figure out how to scale our AI, machine learning, and NLP models,” confirms S&P Global CIO Swamy Kocherlakota. This challenge is widely shared.

Data issues get in the way

Organizational and process rigidity, budgetary constraints, talent shortages and limitations of existing data and technologies are all obstacles to scaling up AI.


More specifically, 72% of respondents “say that data issues are more likely than other factors to jeopardize the achievement of their AI goals by 2025.”

This encompasses the entire life cycle of the data and its quality throughout the processing chain. “Data is one of the biggest challenges we face for [développer l’IA]from data acquisition to data ingestion, data management and quality assurance,” says Rowena Yeo, CIO of Johnson & Johnson.

Scaling requires ticking multiple boxes, as Marks & Spencer CDO Jeremy Pee points out. “Part of the challenge is to build the infrastructure, to build trust in the data, to make it searchable, findable, trusted and well-managed.”

However, this is only one part of the equation to be solved. Scalability must also be introduced at the AI ​​production and design level. “How do you go from a single model to building and supporting hundreds of models? If you don’t fix this part, you just end up creating a lot of inefficiency and frustration. Result: trust begins to break down,” adds the chief data officer.

Perhaps counter-intuitively, the experts surveyed (72% and 92% for the leaders) consider multicloud to be a facilitator of developments in AI – in the same way, however, as open source standards or APIs.

Similar Posts

Leave a Reply

Your email address will not be published. Required fields are marked *