Top 5 AI Predictions From Experts In 2025

The past 12 months have seen majore advances across AI. These include everything from the buildout of massive supercomputers for ongoing AI training to large language models learning how to use computers as well as AI’s expansion into internet search and real-time political polling details.

For this piece of prognostication perspective, five experts were asked to independently share their top three AI projections.

Based on the insights gleaned from this handful of AI experts, 2025 looks to be even more revolutionary for the sector than 2024 with some surprising predictions and trends likely to develop through the end of the new year.

AI Agents Come Of Age

Virtually all the experts mentioned the rise of AI Agents — computer programs that can think, learn and act autonomously — as the most significant trend to watch for in the coming year.

Pascal Bornet is an award-winning expert, keynote speaker and author — his latest book, IRREPLACEABLE: The Art of Standing Out in the Age of Artificial Intelligence, distills his research into practical strategies. He says 2025 will push the AI boundaries beyond our expectations.

“Agentic AI will redefine autonomy. These systems won’t just respond — they’ll take initiative, anticipate needs and act independently, solving problems before we even notice them, becoming indispensable collaborators in both work and life,” Bornet shared in an email.

Ben Torben-Nielsen, Ph.D., MBA, is an internationally recognized AI consultant for businesses with two machine learning patents. He agrees that AI agents hold transformative potential but with a twist — when combined with cheaper small language models.

“To solve a business problem, managers need a dozen tools stitched together. Apart from coding, there is little integration of generative AI into any other workflow. Agentic AI — especially combined with smaller language models — offers a solution by letting AI actually use those tools. This is where the real potential lies,” explained Torben-Nielsen in an email.

While Eduardo Ordax — who has more than 10 years experience in the space focusing on big data, analytics and artificial intelligence working for internationally recognized leaders within the space — predicts AI agents could triple certain gains realized from large language models.

“While current LLMs can help boost productivity by 20-30%, AI agents could potentially double or triple these gains in specific use cases. This shift towards more autonomous AI systems will likely drive broader adoption across industries by streamlining operations and reducing costs,” Ordax stated in a text message.

AI As A Competitive Advantage

Ahmed Banafa, Ph.D., is a technology expert and engineering professor at San Jose State University who shared in an email that generative AI will be safely and securely integrated into enterprise software.

“Businesses will leverage these models for content creation, customer interactions and workflow optimization. This shift is driven by advancements in AI capabilities and their increasing cost-effectiveness, making them crucial for maintaining a competitive edge across various sectors,” explained Banafa.

Julia McCoy, who started her own hybrid AI tech consultancy, First Movers, wrote in a text message exchange that her AI clients are seeing incredible gains — putting them at a distinct advantage over competitors.

“Companies that successfully implement AI-led processes will be able to reduce their workforces by 95%+ while maintaining or increasing output. We’ll see examples of $100M+ companies operating with just two or three people assisted by agentic AI systems. This represents a fundamental reshaping of how businesses operate and where value accrues,” she wrote.

Torben-Nielsen was even more dire and direct for AI dismissers.

“Ignoring AI might put your company out of business. We have seen a similar wave of closures when companies did not adopt e-commerce and it will happen with AI as well. While not an immediate extinction, failing to plan for AI now creates issues down the line. As PwC wrote, ‘Your AI strategy will put you ahead — or make it hard to ever catch up’,” he added.

AI Expenses Will Likely Increase — A Lot

He was also the only expert to forecast a sharp price spike for companies looking to integrate AI in 2025.

“Operating AI is about to get much, much more expensive. First, high-end models will be priced at a premium; OpenAI is experimenting with this and others will follow. Second, while cost per token — the text-based building blocks LLMs use for queries and base pricing on — is plummeting, for other models, increased adoption will cause token usage to skyrocket, which will drive operational costs. Many companies will be facing unexpectedly hefty bills as they likely underestimated operational AI budgets for 2025,” cautioned Torben-Nielsen.

AI Integration With Robots

Bornet and Banafa both mentioned that AI will likely breakout from desktop and mobile devices — moving into mobile robots that we’ll be interacting with much more frequently in the real world.

“Robots will no longer feel futuristic but everyday; wearables and smart devices will merge seamlessly with our environments, making AI a tactile and visible presence. These advancements will tackle ever more complex challenges, from global logistics to personal accessibility, laying the groundwork for a better, more equitable world,” wrote Bornet.

“AI systems capable of processing and integrating diverse data types — text, images, audio and video — will achieve a significantly deeper level of contextual understanding. This will enable more human-like interactions and revolutionize applications in fields such as education, healthcare and entertainment. This progress will be fueled by advancements in computational power and the growing availability of large, multimodal datasets,” expounded Banafa.

Adoption Of Tougher AI Regulations

Banafa was the only expert who discussed the topic of AI regulation — specifically mentioning a regulatory crackdown is likely coming.

“Global AI regulations will become more stringent, emphasizing transparency, ethical use, and accountability. The need to address biases, data privacy and the potential for misuse will compel companies to prioritize compliance alongside responsible innovation. This may somewhat slow unregulated progress but will ultimately ensure safer and more trustworthy AI applications,” he concluded.

AI Consolidation And Shakeouts

While Ordax referenced the likelihood of shifting business models and potential mergers among generative AI providers to remain competitive themselves.

“Currently, training and running LLMs require huge investment in expertise, data collection and compute. To become profitable, many AI labs will transition from being pure LLMs providers to offering software solutions that directly serve end users. This transformation will probably lead to market consolidation, with a few major AI research companies dominating the field,” he concluded.