Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
After another year of rapid development and experimentation with AI, techies and marketing experts think 2025 could help push adoption beyond the testing phase.
The factors at play come from multiple fronts. Tech companies are expanding access to AI content creation, agencies are working on ways to improve workflows for various tools, and enterprise-focused companies are looking for additional ways to achieve better performance for specific applications. Meanwhile, tech companies are rushing to deploy new ways for companies and consumers to use them AI agents.
While there are plenty of topics to watch in the coming year, here are five things industry experts think will happen to AI in 2025 — not to mention all the news from Las Vegas this week. during CES. (Read more in our AI News Timeline 2024 and how platforms are evolving with AI content and ads.)
It’s no surprise that many expect generative AI to continue to disrupt and transform the future of search. Less certain is how quickly, to what extent, and how it will affect competitors, advertisers, and users. It will also be worth watching to see what happens with the ongoing search and ad technology tests – especially when it comes to Google forced to sell my chrome browser.
Agency executives and search professionals expect search to rely less on keywords and more on multimodal capabilities for searching semantic text, images, and videos. Others expect to see more personalization with chatbots and autonomous agents that browse the web. The changes could also lead to brands relying less on clicks and more on impression-based or micro-conversion metrics – especially with AI search platforms with unordered lists, drill-down prompts or deeper chat-based conversations.
Recent research by BrightEdge found that ChatGPT saw a 44% month-over-month increase in traffic, but OpenAI still occupies less than 1% of the total market share. Meanwhile, Google is rapidly increasing the volume of its AI Overviews responses.
Jim Yu, CEO of BrightEdge, said advertisers, publishers and creators will need to rethink their approach to content by optimizing enough to show up in AI results, while finding ways to grow and retain audiences.
“Next year will be a gentle dance for most,” Yu said. “And it will happen industry by industry. Some industries are a little more isolated and come a little later. But you can see e-commerce, publishing and travel starting to hit.”
With the adoption of artificial intelligence, there are still many unresolved risks and unanswered questions: copyright lawsuits, accuracy issues, privacy and bias concerns, and other ethical dilemmas that affect people and professionals.
Companies that adopt AI tools for content and ads also need to make sure they maintain consumer trust. According to Gartner analyst Nicole Greene, this requires being transparent with AI-generated content, listening to audience feedback, and communicating with customers about the benefits and concerns of AI. One Gartner survey found that 80% of consumers think generative AI makes it harder to identify what’s real on the Internet, while 78% said it’s important for AI-generated content to be properly labeled.
“Regulators expect all organizations and their leaders to adhere to responsible use even as regulations change,” Greene said. “For this reason, it is crucial to create a strong foundation for AI governance based on common AI principles now, rather than in the future. This technology is outpacing our ability to regulate it, so organizations must implement safeguards.”
After years of hype around generic big language models, enterprise adoption may depend on smaller, more specialized LLMs to help with industry-specific tasks. One example is artificial intelligence startup Writer, which has developed an industry-specific LLM. While his Palmyra Med model has knowledge and tasks specific to medicine, the Palmyra Fin model focuses more on math, reasoning and calculations. The most a recent introduction is Palmyra Creativewhich debuted tools for creative writing — and creative thinking — in December.
Waseem AlShikh, co-founder and CTO of Writer, thinks enterprise adoption will grow as industry-specific models help improve accuracy and performance based on what companies need for their specific type of business. He pointed out that an accuracy rate of 80% or 90% is not good enough for industries like finance and healthcare, which require maximum accuracy. For him, productivity with AI means paying less money or the same amount of money for better results. But he said some AI models still cost companies more on an hourly basis than they would pay for a full-time employee.
“You need something cheaper, you need high precision. And the way we think we’re going to achieve that for those who embrace it is to have [AI models] that they will develop themselves in such a way that they simply will not make the same mistake again,” AlShikh said. “Because companies that are working on specific gaps aren’t going to change that much if you adopt that without having domain-specific models that can actually learn and have deeper knowledge in those specific areas.”
Joel Burke, a political analyst at Mozilla, thinks open-source AI models will gain momentum as they become simpler, cheaper, safer and more efficient. This could be especially useful for companies that want to protect their IP and user data, as open source models often run locally on the device without sending data to the cloud or third-party providers.
Burke thinks open source will encourage innovation, lower barriers to entry, and give users more choice in choosing an Internet browser or LLM for different tasks. Mozilla is also working on projects like Llamafile, a Mozilla Innovation Project initiative to simplify and make LLM chatbots more transparent.
“When you’re using open source, it’s easier for you to build something,” Burke said. “I think you’re going to see a lot of startups basically take something open source and build on it and use that as their foundation or backbone. Not every startup will have the money for that [other major companies] they have to train their own huge models.”
Shifts in consumer behavior caused by the adoption of AI could also affect how companies think about branding, said David Placek, founder and CEO of naming agency Lexicon. For example, greater adoption of AI voice search—with ChatGPT, Apple’s Siri, and Google’s Gemini—could increase the importance of how brand names sound.
“If ChatGPT gives you five brands to look at, you’re probably not going to look at all five,” Placek said.
Lexicon is also incorporating AI into its processes, such as using LLM to analyze potential brands based on five years of research into sound symbolism and developing tools to organize and search the project archive.
“The understanding of what was said and the memorability of all that mess,” Placek said. “Did something say ‘Siri’ or ‘Sarah’ and stuff like that? is [a name] too soft? As marketers, we will become increasingly sensitive to both the sound and the noise a word makes to make it understandable and better remembered.”