AI Predictions for 2025 from the AI in Business Community
Envisioning: sustainable scalability, reference architectures, ‘agentic AI’ acting autonomously, trust top of mind; wider industry adoption, some reduced headcount, specialized language models
By John P. Desmond, Editor, AI in Business
The community of professionals that read the AI in Business newsletter sent in a range of predictions on the impact AI will have in 2025. We publish a selection here. Happy New Year!
Bob Pette, VP Enterprise Platforms, Nvidia
Seeking sustainable scalability: As enterprises prepare to embrace a new generation of semi autonomous AI agents to enhance various business processes, they’ll focus on creating robust infrastructure, governance and human-like capabilities for effective large-scale deployment. At the same time, AI applications will increasingly use local processing power to enable more sophisticated AI features to run directly on workstations, including thin, lightweight laptops and compact form factors, and improve performance while reducing latency for AI-driven tasks.
Validated reference architectures, which provide guidance on appropriate hardware and software platforms, will become crucial to optimize performance and accelerate AI deployments. These architectures will serve as essential tools for organizations navigating the complex terrain of AI implementation by helping ensure that their investments align with current needs and future technological advancements.
Revolutionizing construction, engineering and design with AI: Expect to see a rise in generative AI models tailored to the construction, engineering and design industries that will boost efficiency and accelerate innovation.
In construction, agentic AI will extract meaning from massive volumes of construction data collected from onsite sensors and cameras, offering insights that lead to more efficient project timelines and budget management.
AI will evaluate reality capture data (lidar, photogrammetry and radiance fields) 24/7 and derive mission-critical insights on quality, safety and compliance — resulting in reduced errors and worksite injuries.
For engineers, predictive physics based on physics-informed neural networks will accelerate flood prediction, structural engineering and computational fluid dynamics for airflow solutions tailored to individual rooms or floors of a building — allowing for faster design iteration.
In design, retrieval-augmented generation will enable compliance early in the design phase by ensuring that information modeling for designing and constructing buildings complies with local building codes. Diffusion AI models will accelerate conceptual design and site planning by enabling architects and designers to combine keyword prompts and rough sketches to generate richly detailed conceptual images for client presentations. That will free up time to focus on research and design.
Percy Liang, cofounder, Together AI and Stanford University associate professor of computer science
"As we move into the new year, I'm excited to see generative AI continue its rapid evolution, especially in areas where progress is already accelerating. Models focused on code and math (anything with well-defined reward signals) will become even more capable, pushing the boundaries of what we can automate and optimize.
I expect open-weight models to reach a level of performance that makes them viable for a wide range of practical applications, making cutting-edge AI more accessible than ever before.
Another area to watch is the growing role of AI-generated audio and video content. We will soon see this kind of content becoming a significant part of our everyday media consumption.
I believe we're on the cusp of a major scientific breakthrough driven by AI, which will have profound implications for research and innovation. The pace of progress in generative AI is only going to accelerate, and I can't wait to see where it takes us next."
Paul Barba, chief scientist, InMoment, offering customer experience services; he was also chief scientist at Lexalytics, an NLP company InMoment acquired in 2021
MLOps and LLMOps combine into AIOps - Machine learning and prompt engineering are very distinct techniques with different tools and use cases. However, they're all part of the broad desire to build robust automation systems. As practitioners get a better handle on the strengths and weaknesses of both, and how they can be used together, expect to see a new wave of frameworks and platforms for leveraging both technologies as one.
LLM quality keeps up its frenetic pace of improvement - There's a lot of chatter that the process of simply scaling LLMs to ever larger data centers may be hitting its last legs. Are we finally hitting the peak of what can be achieved with this technology? My prediction is no. I don't doubt that there are a great deal of brilliant research ideas sitting on the backburner as scaling laws drove the field forward. Expect to see more variation in output as the LLM providers try all different ways of keeping AI advancing, some of which will prove to be enough to see major advances continue.
AI Breaks Down Silos - Businesses have many small AI projects being run out of many departments. The data needed to solve a problem may exist in any number of different silos. Given that everybody is using the same tools, projects will begin organically merging across traditional business boundaries. A lot of focus today is on using AI to access the right information, but even more important than information is accessing the right people. Expect to see thought leadership and case studies appearing where AI brings an organization closer together by providing coordination between groups who otherwise wouldn't speak.
Prakash Arunkundrum, chief operating officer rand general manager, Logitech for Business
AI goes from interesting to mainstream: We blinked, and AI made the jump from a fascinating marvel to securing a place in our everyday lives as consumers. In 2025, AI will firmly be planted in enterprise cases as well, as business and employees become more savvy with these tools to automate their workflows. Companies previously skeptical of AI will embrace its adoption as the tools’ multidimensional value, especially in audio and video, trumps their concerns, particularly in bespoke use cases trained on their proprietary data. With more and more open-source AI and mega players pushing the envelope on cloud-based AI models, companies will start using AI tools as a way of working embedded in their businesses. Globally, all knowledge workers will increasingly use AI tools in everything they do.
Trust will become even more top of mind: Widespread concerns about data privacy, cybersecurity, and lack of transparency will become more important topics. Avoiding another CrowdStrike incident will become top of mind. Fears about AI will be outweighed by the enormous benefits it brings but will simultaneously cast a brighter spotlight on companies’ perceived trustworthiness. Companies that choose to self-regulate, and be transparent about it, will have a competitive edge in an era of “we don’t know who to trust.”
Smart office tools will be non-negotiable keys to employee experience: Regardless of work from home or return to office, next year the speed of technological advancement won’t slow down, and employees will expect the same level of innovation as they enjoy in their consumer tech. In 2025, companies will use sophisticated data analytics to design and refine their hybrid work policies. Metrics like real-time desk and meeting room usage will be used to determine where financial and tech investments should be made for personalized and flexible work arrangements and real estate planning.
Sanja Fidler, VP of AI Research, Nvidia
Predicting unpredictability: Expect to see more models that can learn in the everyday world, helping digital humans, robots and even autonomous cars understand chaotic and sometimes unpredictable situations, using very complex skills with little human intervention.
From the research lab to Wall Street, we’re entering a hype cycle similar to the optimism about autonomous driving 5-7 years ago. It took many years for companies like Waymo and Cruise to deliver a system that works — and it’s still not scalable because the troves of data these companies and others, including Tesla, have collected may be applicable in one region but not another.
With models introduced this year, we can now move more quickly — and with much less capital expense — to use internet-scale data to understand natural language and emulate movements by observing human and other actions. Edge applications like robots, cars and warehouse machinery will quickly learn coordination, dexterity and other skills in order to navigate, adapt and interact with the real world.
Will a robot be able to make coffee and eggs in your kitchen, and then clean up after? Not yet. But it may come sooner than you think.
Industries adopt generative AI: Nearly every industry is poised to use AI to enhance and improve the way people live and play.
Agriculture will use AI to optimize the food chain, improving the delivery of food. For example, AI can be used to predict the greenhouse gas emissions from different crops on individual farms. These analyses can help inform design strategies that help reduce greenhouse gas in supply chains. Meanwhile, AI agents in education will personalize learning experiences, speaking in a person’s native language and asking or answering questions based on level of education in a particular subject.
As next-generation accelerators enter the marketplace, you’ll also see a lot more efficiency in delivering these generative AI applications. By improving the training and efficiency of the models in testing, businesses and startups will see better and faster returns on investment across those applications.
JD Dillon, chief learning officer, Axonify, offering tools for employees who directly interact with customer or the public, “frontline” workers
Artificial intelligence: AI will help companies check training boxes faster and cheaper moving forward. This will result in a reduction in L&D [learning and development] headcount through a combination of process automation and outsourcing to providers using these tools. AI will also help organizations rebalance what people need to know vs what they can access on the job to solve problems and complete tasks.
Labor: Labor hours will continue to tighten, especially if the economy slows down. This reduces L&D opportunities for frontline workers, who are scheduled to the minute and asked to handle extensive task lists with limited staffing. This challenge will get worse if shifts in immigration policies result in labor shortages in certain industries and business categories. The continued introduction of technology, especially automation and AI-enabled tools, may make it more difficult to justify additional investment in frontline training, especially in introductory, high-turnover roles.
Technology consolidation: Organizations will look for opportunities to reduce costs, simplify digital experiences and reduce risk/complexity. Frontline wages and benefits are declining as organizations roll back investments intended to close staffing gaps during the pandemic. This will likely continue, along with tight labor budgeting and reduced hours – unless other shifts cause staffing shortages to emerge in select industries. Collective bargaining will likely take a hit over the next few years because of shifts in the political landscape. When you combine this with the full force return to the office for corporate workers, organizations are re-establishing control over the workplace.
Brian Higgins, chief customer experience officer, Verizon consumer group
“Specialized Language Models (SLMs) are poised to disrupt various industries. Unlike Large Language Models (LLMs) trained on general text data, SLMs are fine-tuned on specific domains, offering enhanced accuracy and efficiency for domain-specific tasks, and they do it in a cost and energy efficient manner. From telecom and healthcare to finance and legal, SLMs will be streamlining processes and improving customer experiences across the board.”
He sees two technologies–agentic AI and retrieval-augmented generation (RAG)--as being on the rise in 2025. Agentic AI involves artificial intelligence and human agents working in tandem to deliver improved outcomes. RAG is a natural language processing technique that combines LLMs with information retrieval systems to produce more accurate and relevant text.
“So, you will see more hyper-personalization, proactive engagement, and seamless integration of technology and human touchpoints. The successful businesses will be the ones that elevate their interactions to meet customer expectations, emphasizing proactive solutions tailored to individual needs. This is especially vital as consumers demand not only efficient but also proactive service across every touchpoint.”
Michael Chime, CEO, cofounder, Prepared 911, and Brad Flanagan, director of its public safety answering point program
Prediction: Technology will reshape how emergency responders operate
Flanagan observes: "It surprises me that emergency responders are excited about technology, because emergency response has been an industry so hesitant to change and where tradition is the foundation. Now, responders want to do their work differently with technology. In recent years, culture was the focus of the industry, now the technology is also in focus.”
Chime predicts: “I think you'll see fewer mistakes from AI technology as models and systems improve at a rapid rate.
One of the hardest jobs in emergency response is listening to a call and taking notes in real time – where technology, like automatic summaries, can step in. You'll see more opportunities in the coming year where vendors will focus on taking these tasks off people's plates — whether it's language translation, QA, or other highly repetitive processes."
