What’s called ‘Generative AI’ (a form of AI that can create content like text, images and even video based on a range of data feeds), in the form of Chat GBT has been around for about 2 years now and enthusiasm is now starting to calm on this area, mainly due to its limitations and costs.
Instead companies want to see proven results of what Generative AI can achieve before installing it throughout these organisations.
Here at Jam we've pulled together the top developments in the pipeline for 2025.
Gen AI hype moves towards proving the benefits
A report by Informa TechTarget’s Enterprise Strategy Group (September 2024) found that despite 90% adoption by organisations, only 8% had truly integrated Generative AI in to their businesses.
A key reason it seems is the variable impacts across job functions; where the AI improves things for one individual but dimishes it for his/her counterpart.
Hence the move by organisations to better understand the benefits overall to the business before fully installing.
Large Language Models move from ChatGBT to reasoning
Organisations are starting to want to see what Large Language Models (LLMs) can drive apart from the origonal chatbots.
Multimodal models are growing in interest, like OpenAI’s text-to-video Sora and ElevenLabs’ AI voice generator, which can handle unstructured nontext data types, such as audio, video and images.
But what about Large Language Models that reason?? It’s December 2024 and Google’s DeepMind revealed an experimental new web-browsing agent called the Mariner. In the middle of a preview demo that the company gave to MIT Technology Review, Mariner got stuck. A product manager at the company had asked the agent to find her a recipe for Christmas cookies that looked like the ones in a photo she’d given it. Mariner found a recipe on the web and started adding the ingredients to Goel’s online grocery basket. But then it got stuck! It couldn’t figure out what flour to use.
Mariner explained its steps in a chat window: “It said, ‘I will use the browser’s Back button to return to the recipe.’” It was remarkable, because instead of hitting a wall, the agent broke the task down into separate actions and picked one that might resolve the problem. And it worked. Mariner went back to the recipe, confirmed the type of flour, and carried on filling Goel’s basket. OpenAI and Google are just the tip of the iceberg, when it comes to building large language models that use similar techniques, making them better at a whole range of tasks. Watch out for more on AI and reasoning!.
Agentic AI models
Agentic AI models are tools designed to autonomously handle tasks for business users, managing workflows and taking care of routine actions, like scheduling and data analysis.
For example tools like Salesforce’s Agentforce which can handle tasks for business users, like managing workflows, routine actions, scheduling and data analysis.
This type of autonomous functionality isn’t new but the way it can adapt and take on new information, in real time, and make decisions is.
AI Robots learning through observation
Instead of processing data, robots can now learn through observing the actions of humans. A bot program called AlphaGo taught itself advanced strategies for playing the game Go, with no training from humans! This is a growing trend of AI – learning independently from daa input and human knowledge.
For Jonathan Siddharth, CEO of Turing, the standout feature for 2025 AI systems will be their ability to learn from human expertise at scale. “The key advancement will come from teaching AI not just what to do, but how to approach problems with the logical reasoning that coding naturally cultivates,” he says.
Open source leading the way
Bill Higgins, VP of watsonx Platform Engineering and Open Innovation at IBM, expects open-source AI models will grow this year.
He cites the high licencing fees preventing companies from seeing a positive ROI, and therefore demand for lower cost solutions. in popularity in 2025. As before organisations are looking more to seeing ROI benefits before widely adopting AI in theie businesses. Cost efficiency and edge data management will become crucial, helping organisations optimise operations while keeping budgets in check.
About us:
Jam is a leading digital marketing agency in the UK with a specialism for website design, UX and AI.
