The 4 biggest AI stories from 2024 and one top prediction for 2025

By all measures, 2024 was the biggest year yet for artificial intelligence – at least when it comes to the commercialization of the technology.

The large language model (LLM) boom fueled by the launch of ChatGPT in late 2022 has shown no signs of slowing down, with several new LLMs being introduced not only by OpenAI and tech giants strong like Microsoft, Meta and Google, but also scores of other startups. and individual developers.

Reports of slowing AI research have been, if not unfounded, certainly exaggerated for now.

In addition, new technologies emerged outside of the Transformer architecture that underpins most large LLMs, such as Liquid AI Foundation Models.

And finally, companies began to take the “active” approach to AI – developing specialized AI-powered bots, applications, and workflows that can work on specific problems independently, or with less human stewardship than the usual back and forth of LLM chatbots. .

Narrowing the year's news down to a top 14 was a daunting endeavor, much less a top 10 or top 4. But I've gone ahead and tried, even though I cheats a bit by combining several stories into larger themes. In my eyes, here's what will make the biggest impact moving out this year:

1. Expand OpenAI far and wide beyond ChatGPT

Arguably the company most responsible for the gen AI era did not miss a beat this year, despite increased competition from newbies and legacy technology, even its investor and partner Microsoft himself.

o1 Model: OpenAI released the first new family of general purpose models beyond its GPT series, the “reasoning” sequence o1which allocates more time to process complex proposals, leading to higher accuracy. It is particularly effective in science, coding and reasoning tasks.

Model o3: It followed the o1 model from September with a year-end announcement in the evening more advanced o3 model. While this won't be available publicly or even to third parties until early 2025, it shows that OpenAI isn't resting on its laurels.

Search ChatGPT: This feature, initially launched as a stand-alone invite-only product is called Search GPT before being dropped into ChatGPT proper, enabling more real-time web information retrieval within ChatGPT and a refined display of search results, increasing its utility for up-to-date queries and forward -face-to-face with Google, Bing, and modern Perplexity.

Canvas: Introduced in October, Canvas extends the ChatGPT interface beyond the chat interface to a workspace like pane that can dynamically update content at the user's request, such as editing a document or coding project. In fact, it was hard not to see it as a response to, or at least a comparable feature to Anthropic's Artifacts was announced several months earlier.

Sora: After nearly a year of teasing us with its under-the-hood video generator module, OpenAI in early December launched Sora to great effectquickly inviting a wide range of ideas as it sought to differentiate itself in the highly competitive AI video space with a unique and well-thought-out feature interface and storyboard .

2. Open source AI built off of it

Lama 3 and 3.1: Meta included Lama 3 in Aprilsetting a new standard for performance in open source AI, then quickly followed with Llama 3.1 in July with 405 billion parameters. Versions of Llama 3.1 were used to power Meta AI, the company's assistant integrated across platforms such as WhatsApp, Messenger, Instagram, and Facebook, aiming to be the most widespread AI assistant.

Lama 3.3: Released in December 2024, Llama 3.3 delivered performance comparable to larger models but at a fraction of the computing cost, making it more affordable for enterprise applications.

At the same time, Chinese models such as Alibaba's Qwen-2.5 family and DeepSeek has the new V2.5.5 and R1-Lite preview seemingly out of nowhere to top some benchmark charts, and Nvidia itself went beyond providing graphics cards and software architecture to launch its own open source, powerful Model Nemotron-70B.

Nous Research, a small outfit in San Francisco aims to offer more personal and less restrictive AI models as open source, also appeared several cool new thoughts.

And let's not forget France Mistralwhich quickly expanded its own open source and proprietary AI offerings.

3. Google's Gemini series became a leading contender for the best available

In the comeback story of the year, Google's Gemini series of AI models that were once derided for their weird image generation and criticized for being too “woke” came back with new, more powerful versions that -now tops the third-party performance benchmark charts. and are increasingly attractive to developers and businesses.

Google introduced Gemini 2.0 Flasha multimodal AI module that supports streaming video analysis and sees and learns what you're doing on your screen, and follows it with Gemini 2.0 Flash Thinking which competes with OpenAI's o1 and o3 reasoning models.

4. The AI ​​agent swept the campaign

As the year progressed, “active” AI went from being a fad to a real series of major product announcements and initiatives by major enterprise software vendors. Take for example:

Salesforce 2.0 Executive: Salesforce released Agentforce 2.0 a few days ago, an advanced AI agent program to enhance reasoning, integration and customization features across its CRM and sales offerings, as well as Slabgreatly improving enterprise productivity tools.

Joule at SAP: SAP turned the Joule chatbot into An AI agent powered by large open language modules (LLMs), driving innovation and efficiency in enterprise settings.

Google's Astra project: As part of the Gemini 2.0 initiative, Google launched Project Astra, an AI assistant designed to provide real-time, contextual answers by leveraging Google's suite of services, focused on productivity users and enhance decision making.

My big prediction for 2025: AI-generated content will reign supreme

Building on these advances, 2025 is set to see a proliferation of AI-generated content across business and consumer domains, notably with everyone from OpenAI to Meta, Google, Microsoft, Apple, even Elon Musk's xAI now has AI image generators built into their offerings.

This extension streamlines content creation, enhances personalization, and drives efficiency in multiple areas.

Additionally, we anticipate widespread use of large-scale language models (LLMs) and AI-powered generative robots in commercial and consumer settings, revolutionizing automation and human-robot interactions.

That's all in the last #AIBeat newsletter for 2024. Thanks for reading, writing, subscribing, sharing, commenting, and being here with us. Looking forward to sharing more and hearing more from you all in 2025.

Happy Holidays and a Happy New Year from all of us at VentureBeat to you and your loved ones.