The year 2024 has been pivotal in showcasing the transformative potential of generative artificial intelligence (AI), following a 2023 that placed the technology within mainstream tech discussions. What began as a novel text-based chatbot trend has now evolved into essential components of various tech products, with AI demonstrating practical applications across music, video generation, and even autonomous capabilities. The anticipated AI bubble showed no signs of bursting this year.
This year has also seen the introduction of large language models (LLMs) focusing on enhanced reasoning, signaling the dawn of AI PCs—often referred to by Microsoft as Copilot+ PCs. The open-source AI sector experienced accelerated growth, among other significant developments that have made headlines throughout 2024. Here’s a look at the key milestones that have shaped the AI landscape this year.
OpenAI’s Year of High-Performance AI Models
While OpenAI kickstarted the generative AI movement with its Generative Pre-trained Transformer (GPT) architecture in late 2022, the tech giants quickly entered the competitive fray by the end of 2023. Companies like Google, Microsoft, Meta, and Amazon released various AI models, vying for top benchmark performance.
OpenAI launched its advanced reasoning-focused GPT-4o AI model in May 2024, followed by the GPT-4o Mini in July. The year concluded with the release of the complete o1 model and the highly anticipated text-to-video model, Sora.
The company further enhanced the ChatGPT app with its Advanced Voice Mode, offering users new interaction methods. Additionally, OpenAI introduced its own search engine, ChatGPT Search, directly integrated into the chatbot platform.
A significant breakthrough for OpenAI occurred with its partnership with Apple, allowing for the integration of ChatGPT into Apple’s Intelligence tools. Following this collaboration, OpenAI unveiled standalone macOS and Windows applications for ChatGPT.
Google’s Diverse Set of AI Offerings
Google made waves with numerous model releases throughout the year. In February, it launched the Gemini 1.5 series, which includes the Gemini 1.5 Pro boasting one trillion parameters. The year wrapped up with the introduction of the Gemini 2.0 series, featuring the Flash model available for public preview and a larger model exclusive to paid subscribers.
Google DeepMind also unveiled several groundbreaking developments, including the Imagen 3 model for image generation and the Veo 2 model for video creation. The music generation model, MusicLM, was previewed. The company also introduced NotebookLM, an AI tool to process large documents that can create engaging podcasts with two virtual hosts.
Moreover, Google enhanced the Gemini offerings by adding a two-way voice communication feature called Gemini Live and integrating the Gemini AI assistant across most Google Workspace applications, such as Gmail, Docs, Slides, and Sheets.
Meta, traditionally associated with social media, showcased its prowess in AI by launching several small language models (SLMs), many of which were made available as open-source. The company introduced its Large Language Model Meta AI (Llama) series, including the 70B and 30B coding-focused models, and the largest open-source model, Llama 3.1 405B, along with multiple instruct models. The expansion of Meta AI into Facebook Messenger, Instagram, and WhatsApp hit several regions, including India, before reaching a global audience by September.
The AI-powered chatbot was also integrated into Ray-Ban Meta glasses, offering real-time vision processing capabilities.
Microsoft and the Era of Copilot+ PCs
Microsoft carved a niche in the AI sector by integrating AI models from OpenAI while also debuting its own innovations in the PC landscape. The company announced a partnership with Snapdragon and later collaborated with Intel and AMD to introduce the AI PC classification, introducing a physical Copilot button on keyboards—a hallmark of the Copilot+ PC era.
The integration of Copilot tools across Microsoft 365 products and the addition of voice and vision capabilities to the chatbot marked significant advancements. Furthermore, Microsoft launched the AI-powered Recall feature, currently in beta, enabling users to ask questions related to past activity on their devices.
Amazon’s Role as AI Aggregator
Despite perceptions that Amazon was slow to enter the AI domain, the company adopted a distinctive approach to stay relevant in 2024. While not launching standout products, the company introduced the Rufus AI tool within the Amazon app, which serves as a shopping assistant, and unveiled the Titan series of AI models alongside a video generation model for enterprises.
In a strategic move, Amazon emerged as an AI aggregator, incorporating models from numerous third-party providers into its Amazon Web Services (AWS) platform. Additionally, the company worked on tools to enhance response efficiency and mitigate hallucinations in AI interactions while upgrading its servers to support extensive AI processing.
Other Notable AI Announcements
Smaller AI companies also made significant strides in 2024. Anthropic continued its success with the Claude AI series, releasing Claude 3 earlier in the year and Claude 3.5 by year’s end. A beta desktop app for Mac and Windows, along with standalone mobile apps, further augmented Claude’s capabilities, including improved tool use and PDF understanding.
The AI search engine Perplexity introduced a Pro mode for more complex queries and launched a standalone Mac app, although its move to include ads even for premium subscribers drew some backlash.
Mistral maintained its focus on fully open-source AI models, releasing the 8x22B Mixture of Experts (MoE) models and the Mixtral Open 2 LLM, along with the innovative Pixtral 12B AI model boasting computer vision capabilities.
AI in 2025: A Brief Outlook
While this overview captures significant events in the AI sector throughout 2024, it is impossible to mention every noteworthy release given the rapid pace of innovation. Looking ahead to 2025, the momentum for AI technology is expected to continue unabated.
The upcoming year is anticipated to witness the rise of agentic AI, enabling seamless integration into various platforms and devices. This advancement would allow users to, for example, instruct a chatbot to purchase tickets or find the best deals without further intervention.
Furthermore, enhancements in memory capabilities for chatbots are expected, moving beyond basic retrieval-augmented generation (RAG) to provide more effective user assistance. The accessibility of real-time video processing is also anticipated to improve, and India is expected to make significant advancements in AI adoption during 2025.
