3 Big AI Trends to Watch in 2024
2023 was a transformative year for generative AI, as groundbreaking tools like ChatGPT and Microsoft Copilot brought this technology from research labs into the daily lives of millions. In 2024, we anticipate AI becoming even more accessible, nuanced, and deeply integrated into technologies aimed at simplifying daily tasks and addressing global challenges. Here are three pivotal AI trends to watch this year.
1. Small Language Models (SLMs)
You may already be familiar with the immense capabilities of large language models (LLMs), which power tools like Copilot. These models, though highly effective, demand significant computational resources. Enter small language models (SLMs) — compact yet powerful versions of LLMs that promise to democratize AI access.
SLMs contain billions of parameters—a stark contrast to the hundreds of billions found in LLMs—but are optimized to run offline on devices like smartphones. These models are more cost-effective, making advanced AI accessible to a broader audience.
“Small language models can make AI more accessible due to their size and affordability,” says Sebastien Bubeck, head of the Machine Learning Foundations group at Microsoft Research. Recent advancements, such as Microsoft’s Phi and Orca models, demonstrate that SLMs can rival or even surpass LLMs in certain applications.
By leveraging curated, high-quality training datasets, SLMs challenge the conventional belief that larger models are inherently superior. Expect 2024 to bring further innovations in SLMs, expanding their role in research and practical applications.
2. Multimodal AI
While most LLMs process a single data type, such as text, multimodal models are revolutionizing the AI landscape by seamlessly integrating multiple data types, including text, images, audio, and video. This innovation enhances tools across search, creativity, and accessibility platforms.
For instance, Copilot’s multimodal capabilities allow users to upload images and receive contextual insights, such as the historical significance of a monument. Microsoft Designer leverages multimodal AI to generate images based on user descriptions, while custom neural voices enhance accessibility tools by creating natural-sounding speech for individuals with vocal impairments.
“Multimodality has the power to create more human-like experiences,” says Jennifer Marsman, principal engineer at Microsoft. This approach mirrors human communication, which relies on sight, sound, and speech, making interactions with AI more intuitive and effective.
3. AI in Science
AI is poised to accelerate breakthroughs in science, addressing critical global challenges such as climate change, energy crises, and healthcare. By harnessing the power of AI, researchers are transforming traditional scientific processes.
For example, Microsoft researchers are developing AI tools for sustainable agriculture, including weather prediction systems and carbon estimators, to support climate resilience. Chatbots equipped with farm-specific data can assist farmers by identifying weeds or optimizing irrigation strategies.
In healthcare, AI is being used to combat diseases, develop new drugs, and discover innovative molecules for life-saving medicines. These advancements drastically reduce the time required for trial-and-error experimentation from years to mere weeks.
AI is also redefining materials science. Recent breakthroughs highlight how AI-powered tools and high-performance computing can identify safer and more efficient materials, such as less toxic components for batteries.
“Artificial intelligence is driving a revolution in scientific discovery,” says Chris Bishop, director of Microsoft Research’s AI4Science team. This revolution represents not just incremental progress but a transformative shift in how we approach global challenges.