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

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

Reports of a slowdown in AI research have proven to be, if not unfounded, certainly overstated by now.

Additionally, new technologies beyond the Transformer architecture that underpins most major LLMs are beginning to emerge, such as Liquid Foundation Models of Liquid AI.

And finally, companies are starting to fully embrace the “agent” approach to AI – developing specific AI powered bots, applications, and workflows that can work on specific problems independently, or with less human stewardship rather than the typical back-and-forth of LLM chatbots. .

Distilling the year’s news into a top 14, let alone a top 10 or top 4, is a daunting endeavor. But I went ahead and tried, even cheating a little by combining several stories into larger themes. In my view, here are the biggest impacts from this year:

1. OpenAI has expanded far and wide beyond ChatGPT

The company probably most responsible for the start of the gen AI era has not missed a beat this year, despite the intensifying competition from new and legacy tech, even its own investor and partner Microsoft.

o1 model: OpenAI releases the first new family of large general purpose models beyond its GPT series, the o1 “reasoning” serieswhich takes more time to process complex stimuli, resulting in higher accuracy. It is especially effective in science, coding, and reasoning tasks.

o3 model: It follows the o1 model from September with a blockbuster year-end announcement of an even more advanced o3 model. Although it will not be available to the public or even to any third parties until early 2025, it shows that OpenAI is not resting on its laurels.

ChatGPT search: This feature, initially launched as an invitation-only stand alone product called SearchGPT before collapsing into ChatGPT precisely, can make more real-time web information to retrieve the content of ChatGPT and a refined presentation of search results, improving its use for up-to-date question and head-to-head against Google, Bing, and the new Perplexity.

Canvas: Introduced in October, Canvas extends the ChatGPT interface beyond a conversation to a workstation like pane that can update content dynamically at the user’s request, such as editing a document or coding project. Of course, it’s hard not to see it as a reaction to, or at least an equal part of Anthropic’s Artifacts was announced a few months ago.

tod: After nearly a year of teasing us with his carefully guarded video generator model, OpenAI in early December finally launched Sora to the massesquickly invited a wide range of reactions as it sought to differentiate itself in a hotly competitive AI video space with a unique and well-thought-out interface and storyboarding feature.

2. Open source AI takes over

Llama 3 and 3.1: Meta is introduced Llama 3 of Aprilset a new standard for open source AI performance, then quickly followed it up with Llama 3.1 in July with 405 billion parameters. The Llama 3.1 versions are powered by Meta AI, the company’s assistant integrated into platforms such as WhatsApp, Messenger, Instagram, and Facebook, aiming to become the most widely used AI assistant.

Call 3.3: Released in December 2024, Llama 3.3 provides performance comparable to larger models but at a fraction of the computational cost, making it more accessible for enterprise applications.

Meanwhile, Chinese models like Alibaba’s Qwen-2.5 family and The new V2.5 of DeepSeek and R1-Lite Preview appears nowhere to top some of the benchmark charts, and Nvidia itself is more than supplying graphics cards and software architectures to launch its own open source, powerful. Nemotron-70B model.

Nous Research, a small San Francisco outfit aims to offer more personal and less restrictive AI models as open source, also debuted some COLD NEW ideas.

And let’s not forget France Mistralwhich is rapidly expanding its own open source and proprietary AI offerings.

3. The Google Gemini series has become a serious 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” are back roaring with new, more powerful versions that now top third-party benchmark performance charts. and more attractive to developers and businesses.

Introduced by Google Gemini 2.0 Flasha multimodal AI model that supports streaming video analysis and can see and teach what you are doing on your screen, and follow it in Gemini 2.0 Flash Thinking which competes with OpenAI’s o1 and o3 reasoning models.

4. Agentic AI is sweeping the business

Over the course of the year, “agent” AI has moved from buzzworld to a veritable series of major product announcements and initiatives by leading enterprise software vendors. Take for example:

Salesforce’s Agentforce 2.0: Salesforce unveils Agentforce 2.0 a few days ago, an advanced AI agent program to enhance reasoning, integration, and customization features across CRM and sales offerings, as well as Slacksignificantly improving business productivity tools.

The Joule in SAP: SAP converted its Joule chatbot into a The AI ​​agent is powered by open-source large language models (LLMs), driving innovation and efficiency in business settings.

Google’s Project Astra: As part of the Gemini 2.0 initiative, Google launched Project Astra, an AI assistant designed to provide real-time, contextual answers by using a suite of Google services, aimed at improving the user productivity and decision-making.

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

Building on these developments, 2025 is poised to witness an influx of AI-generated content in business and consumer domains, especially with everything from OpenAI to Meta, Google, Microsoft, Apple, etc. Elon Musk’s xAI now has AI image generators built on their offerings.

This expansion will streamline content creation, improve personalization, and drive efficiency across sectors.

Additionally, we anticipate the initial large-scale deployment of large-scale language models (LLMs) and generative AI-powered robotics in commercial and consumer settings, revolutionizing automation and human-robot interaction.

That’s it for the last #AIBeat newsletter for 2024. Thanks for reading, writing, subscribing, sharing, commenting, and joining 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.