Looking Ahead: 5 AI Trends to Watch Out for in 2024/25

2023 was the year of rapid development in AI - however, the growth of AI shows no signs of slowing down in 2024. As of June 2024, the global AI market size has already surpassed $184 billion, adding around $50 billion since 2023. It’s poised to reach an astounding $826.73 billion by 2030. According to Goldman Sachs, global AI investment is likely to reach $200 billion by 2025.

As AI continues to grow, the focus is now shifting from experimentation to mainstream business applications. Here are the 5 key AI trends in 2024/25 which reflect this shift.

Multimodal AI

At EmTech MIT event in November 2023, Mark Chen, head of frontiers research at OpenAI, said:

"We want our models to see what we see and hear what we hear, and we want them to also generate content that appeals to more than one of our senses."

Most of today’s leading AI tools like ChatGPT are text-based. However, this is going to change soon.

Imagine this: you click a photo of your fridge’s interior and input a voice command to an AI app. In seconds, it whips up delicious recipe ideas based on what you have inside the fridge, complete with a detailed written recipe. It's like having a personal chef at your fingertips!

This is what multimodal AI is all about.

It’s a type of AI system that can handle inputs in varied formats such as audio, images, text, videos, and standard numerical data. Therefore, multimodality is the capability of an AI system to process these different data formats simultaneously and gain a deeper understanding of the context behind the information.

Multimodality will help AI models have a better understanding of the natural world. Currently, the large models are trained mainly with text data - however, multimodality will open the door for a gigantic amount of new training data. In the words of Cohen:

 “As our models get better and better at modeling language and start to hit the limits of what they can learn from language, we want to provide the models with raw inputs from the world so that they can perceive the world on their own and draw their own inferences from things like video or audio data,"

Multimodality is not just experimentation—it has several real-world applications in healthcare, customer service, security, surveillance, and entertainment. For instance, in healthcare multimodal AI tools can analyze medical images or data from health trackers such as smartwatches to diagnose underlying health issues, if any.

Progress toward autonomous AI agents

According to computer scientist Peter Norvig, a fellow at Stanford's Human-Centered AI Institute,

"2023 was the year of being able to chat with an AI…but the interaction was always you type something in and it types something back. In 2024, we'll see the ability of agents to get stuff done for you. Make reservations, plan a trip, and connect to other services.”

Peter Norvig is essentially indicating a natural progress towards more autonomous AI systems, also called agentic AI. Unlike the AI tools of today, these AI tools won't work based on prompts or predetermined programs. Rather, these tools will be proactive and capable of independent decision-making—for example—managing investment portfolios based on the changing market conditions.

Agentic AI systems show goal-driven behavior and flexible decision-making. They are powered by sophisticated algorithms and sensory data to perform actions in real time, continually learning and improving their performance through ongoing feedback.

Agentic AI has a wide range of applications ranging from robotics and self-driving cars to smart personal assistants and complex simulation environments. In these settings, they can autonomously handle tasks, adapt to changing conditions, and efficiently meet specific goals.

According to Gokul Rajaram, Product and Business Helper at DoorDash:

“The next big thing in 2024 will be an explosion in AI agents of all kinds, focusing on every consumer need and on every kind of business transaction."

Increased Customization

The AI tools of the future are set to be more geared toward the specific needs of an industry vertical, business function, or company. While large language models (LLMs) are versatile and can handle a wide range of tasks, they may not be suitable in situations where specialized knowledge is required to address specific business challenges (especially in fields like healthcare, finance, or legal).

In comparison, domain-specific AI solutions can provide more targeted expertise and better performance for particular applications. These customized enterprise AI models can meet niche business and customer requirements.

More and more businesses are choosing customized AI solutions that meet their specific needs, data environments, and unique challenges. This shift towards personalized AI means we are going to see more specific-purpose and niche models in the coming years.

Open source AI Trends

In July 2024, Mark Zuckerberg, Chairman and CEO of Meta, wrote a blog post titled “Open Source AI Is the Path Forward”. In the post, he argues that open-source models are beneficial for both developers and users as they can easily train, fine-tune, and distill their own models. According to Yann LeCun, a leading voice in the AI field, “Open Source AI models will soon become unbeatable. Period.”

Open source is fueling the next generation of AI innovations. A host of platforms like GitHub and Hugging Face enable open-source AI applications to beat the established closed platforms like OpenAI in their own game. For instance, last year, GitHub experienced a 148% year-over-year increase in contributors and a 248% year-over-year surge in the total number of generative AI projects.

One of the key drivers of open-source AI's growth is that it democratizes AI applications by lowering barriers to entry for businesses and individuals. Traditional AI development often requires a humongous amount of financial investment in proprietary software and specialized talent. However, with open-source alternatives like TensorFlow, PyTorch, and Hugging Face, even small startups and independent developers can build and deploy sophisticated AI models.

This accessibility not only empowers innovation but also encourages a diverse range of applications, from healthcare to finance, education, and beyond. As a result, we are witnessing a surge in creative use cases that address real-world problems.

Evolving AI regulations

In the last 2 years, developments in I have been much more rapid than AI regulations could catch up. This has resulted in a situation in which existing regulatory frameworks are not able to address the risks and concerns around AI. Given this background, we are likely to see more comprehensive AI regulations in the next 1-2 years.

The EU AI Act, published in 2023 was the first major regulation on artificial intelligence. It analyzes and classifies various AI applications based on their potential risks and subjects these applications to different levels of regulations, accordingly.

In the US, Silicon Valley's epicenter California has recently taken the first step to regulate artificial intelligence with a new AI safety bill, SB 1047. It would impose various safety restrictions and requirements on advanced AI models, to allow California to benefit from the technology while avoiding the most severe risks.

While AI is a rapidly developing field, watch these major trends to align your businesses' strategies to reap maximum strategic advantages.