As a leading data analytics and research company, staying at the forefront of artificial intelligence (AI) advancements is integral to our mission at Decimal Point Analytics (DPA). The AI landscape is evolving rapidly, so understanding the key trends that shaped 2024, and anticipating what’s on the horizon for 2025 is critical for us to help our clients take the next steps on their transformation journey.
Key AI Trends That Shaped 2024
Proliferation of Generative AI Applications: The year saw a surge in generative AI tools across industries, facilitating the creation of content, designs, and solutions, as well as streamlining workflows and fostering innovation. The momentum is so strong, that traditional machine learning (ML) algorithms, and even neural network Generative implementations appear to be losing steam. As per a recent Microsoft-IDC report:
- Al usage jumped from 55% of companies to 2023 to 75% in 2024.
- For every $1 a company invests in generative Al, the ROI is $3.7x, the top leaders using generative Al are realizing an ROI of $10.3.
- 43% say productivity use cases have provided the greatest ROI.
- The ROI of generative AI is highest in Financial Services, followed by Media & Telco, Mobility, Retail & Consumer
- Packaged Goods, Energy, Manufacturing, Healthcare and Education.
- On average, Al deployments are taking less than 8 months and organizations are realizing value within 13 months.
Enhanced Productivity: The primary business outcome that companies are trying to achieve with AI is enhanced productivity. The aforementioned study reveals that 92% of AI users surveyed are using AI for productivity, and 43% say productivity use cases have provided the greatest ROI. In the initial low-hanging use cases, employees are saving 15 to 30 minutes a day using Copilot for tasks such as summarizing chats, generating presentations and building executive summaries.
“In 2024, Generative AI reshaped industries with unmatched momentum, delivering up to 10.3x ROI in sectors like financial services, media, and mobility. As we move into 2025, the focus will shift from productivity gains to advanced, custom-built AI applications, unlocking even greater potential across industries.”
Multimodal AI Integration: In 2024, AI systems capable of processing diverse data types—such as text, images, video and audio—became more prevalent. This advancement has enabled more intuitive and versatile AI applications, enhancing user interactions across various platforms.
-
Advancements in AI-Driven Healthcare: AI technologies made significant strides in healthcare – improving business management, diagnostics, and patient care through predictive analytics and efficient data processing.
-
Explainable AI: Throughout the year we have witnessed a growing preference among clients for AI models that are transparent, and not opaque. There is a clear preference for models that provide clarity, offer insights into the factors and explain the logic behind the AI’s decision. This is especially relevant in the financial services industry where regulatory compliance demands transparent AI models. Even hedge funds have expressed a demand for explainable AI solutions.
What lies ahead in 2025 and beyond
- The AI Native Opportunity: The past witnessed two monumental shifts in software technology and business models. The first shift, driven by the PC (personal computer) revolution, led to development of distributed client-server architecture, and was characterized by substantial license sales and annual maintenance contracts. The second shift was driven by the internet revolution, leading to the emergence of familiar web-based SaaS applications with monthly subscription models.
With the Gen AI revolution, 2025 may now witness the third significant shift in software architecture. If we consider the entire AI stack it resembles the chart below (courtesy Andrew Ng)
While the initial excitement has been around the cloud infrastructure and foundation models, the real opportunity, perhaps, lies in the application layer. Rapidly emerging innovative applications that leverage the AI infrastructure and foundation models, threaten to replace well-established SaaS applications. The potential market size for such AI-based applications is expected to be substantially large, as they offer greater business benefits compared to traditional applications. Some estimates even suggest that the AI applications market could be ten times larger than SaaS market, reaching US$300bn versus US$30bn.
Shift from Productivity to More Complex Use Cases: Moving forward, organizations will likely shift their focus from productivity use cases to functional and industry-specific use cases. We expect that over the next five years or so most organizations would invest in advanced custom-built AI solutions.
Agentic AI: A pivotal technology advancement driving this transformation is the development of Agentic AI system – systems that can act autonomously to achieve specific goals. These systems function much like humans, reflecting on requirements, conducting research, and critiquing their own work.
Such agentic workflows can improve the outcome of AI systems dramatically. For instance, consider software coding. On a standardized test for coding GPT 3.5 scores about 53%, while, with a 10x increase in number of parameters, GPT 4 scores about 69%. On the other hand, agentic systems utilizing GPT 3.5 score above 90% on this test.
Agentic AI can adapt to new information and changing environments, making it particularly valuable in dynamic and unpredictable situations. Autonomous agents could potentially manage multiple tasks simultaneously across various domains, from industrial automation to personalized customer service.
- Human in the Loop: Given the current state of the technology, in our opinion human in the loop system design – where a human validates the actions that the AI agent recommends – is still preferable in most scenarios and is a must for critical systems.
