Every single year on December 31st, many people reflect on the past 365 days. “What has changed since one year ago?”, “Have I learned anything new?”, and “What major events have occurred over the past year?” are all questions that one might ask. Today, I want to ask a specific question: “How has AI evolved over the past year?” Specifically, I want to delve into discoveries, innovations, and progress in the field. By reflecting on the evolution of AI in 2024, we not only get informed on recent trends, but also get to make predictions on what entails the AI industry in 2025. So, without wasting any more time, let’s dive in.
LLM to LMM
If you were to tell someone in 2023 that they could not only input text into ChatGPT but also images and audio, they would be baffled. Okay, maybe not BAFFLED, but they would be pleasantly surprised. We now live in that reality where, at least with gpt-4o, we can input multiple types of data (images, videos, raw text, structured data, etc). While the idea of multimodal models is not unique to 2024 and the first multimodal model (GPT-4) was created in 2023, they have definitely become a lot more widespread and talked about over the past year. Besides recent versions of GPT, other multimodal models such as Gemini, models from OpenAI such as DALL-E, Pixtral 12B from Mistral and many more have come out in the past year.
AI in Education
The question of how AI should be used in a school environment has become large in the past year. Many students have been caught “cheating” by using AI models such as ChatGPT, Gemini, Copilot, and Claude to complete assignments and homework. On the other hand, many teachers find AI useful for grading assignments and assessments. In fact, one of my own teachers admitted that they had seen one of their coworkers using ChatGPT to write a letter of recommendation for one of their students. Overall, AI seems to have many use cases in the field of education, some being good while others are bad. Creating rules and regulations for the usage of AI in this industry is definitely a must for upcoming years, especially seeing the events that have occurred over 2024. Even though this issue is not unique to 2024, the question has definitely drawn many concerns and reached its peak this year. It seems that this issue will continue to grow in the future.
AI in Entertainment
AI in the entertainment industry has also become a large point of controversy. On one hand, AI can lead to the production of media such as video games, movies, and TV shows being much faster and efficient. On the other hand, many claim that the usage of AI in these pieces of media will lead to less creativity, innovation, and individuality and instead prompts automation and a loss of human originality. Most companies will likely prefer to use AI in their work soon as it will (presumably) cost less than hiring human workers. The success of using AI will depend on how consumers react to its usage in the pieces of media that they purchase. In the end though, there are many different perspectives to this issue and many different angles you could approach it from. There is no right or wrong answer, so we’ll have to wait and see for any developments in 2025 and beyond.
Besides LLMs
AI is not just Large Language Models, so let’s look at some developments in other areas such as computer vision, general NLP, and more.
Computer Vision Breakthroughs
2024 saw remarkable advances in computer vision technology. Models like Stable Diffusion 3 and DALL-E 3 made leaps in image generation quality. We also saw large improvements in real-world applications like medical imaging diagnosis, autonomous vehicle perception, and industrial quality control. Computer vision has become increasingly integrated into everyday applications, from smartphone cameras to security systems.
Robotics and AI
The robotics field made substantial progress in 2024. Companies like Figure, Agility Robotics, and Boston Dynamics have shown more capable and versatile robots. We saw improvements in robot dexterity, spatial awareness, and the ability to perform complex tasks. The integration of AI with robotics created more intuitive human-robot interactions, opening new possibilities in manufacturing, healthcare, and home assistance.
AI in Scientific Discovery
One of the most promising developments has been AI’s growing role in scientific research. Tools like AlphaFold continued to advance our understanding of protein structures, while new AI models helped accelerate drug discovery and materials science research. The combination of AI with quantum computing also showed promising results in solving complex scientific problems.
What the future is going to look like
Overall, AI has grown more in the past few years than ever before, and this will likely continue into 2025. More accurate and efficient autonomous driving, diagnostic, and other algorithms will be introduced. We can expect that AI will also become more integrated into our everyday lives. We will also have to come up with laws and regulations in response to the rapid evolution of AI.
Finally, I want to wish everyone a happy new year. I hope that you have an amazing 2025, whether it be mentally, physically, socially, or financially. Thank you!.