The 4 biggest AI stories from 2024 and 1 major prediction for 2025

By all accounts, 2024 has been the biggest year for AI yet, at least when it comes to commercializing the technology.

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

Reports of AI research slowing down have proven to be, if not unfounded, then certainly exaggerated at the moment.

In addition, new technologies that go beyond the transformer architecture that supports most large LLM programs are beginning to emerge, e.g Liquid foundation models for liquid AI.

Finally, companies are beginning to fully embrace the “agent” approach to AI – developing specific AI-powered bots, applications and workflows that can work on specific problems autonomously, or with less human supervision than traditional LLM chatbots. .

Distilling the year’s news stories down to the top 14, let alone the top 10 or top 4, was a bewildering effort. But I went ahead and tried, albeit a little cheating, by combining several stories into larger themes. In my opinion, here’s what will make the biggest impact after leaving this year:

1. OpenAI has expanded widely beyond ChatGPT

The company arguably most responsible for launching the era of artificial intelligence hasn’t missed a beat this year, despite intense competition from upstarts, legacy technology, and even its investor and partner Microsoft.

Model o1: OpenAI has released the first new family of general-purpose large-scale models beyond its GPT series, “Inference” series o1which allocates more time to processing complex claims, resulting in higher accuracy. It is especially effective for science, programming, and reasoning tasks.

o3 model: He followed the o1 model from September with a brilliant year-end announcement of a tie More advanced O3 model. Although this won’t be available to the public or even any third party until early 2025, it shows that OpenAI is not resting on its laurels.

Search ChatGPT: This feature was initially launched as a standalone invite-only product called SearchGPT Before being properly integrated into ChatGPT, it allows for more real-time web information retrieval within ChatGPT and improved presentation of search results, enhancing its utility for modern queries and competing head-to-head with Google, Bing, and newcomer Perplexity.

fabric: Submitted in October, fabric It extends the ChatGPT interface beyond a conversational interface into a workstation like part that can dynamically update content based on user request, such as editing a document or coding project. Of course, it was hard not to see it as a throwback, or at least as a comparable feature The human artifacts were announced several months ago.

Sora: Nearly a year after we teased a closely guarded video generator prototype, In early December, OpenAI finally released Sora to the massesquickly invited a wide range of reactions as it sought to differentiate in a highly competitive AI-powered video space with a unique, well-thought-out interface and storyboarding feature.

2. Open source AI has taken off

Llamas 3 and 3.1: dead foot Llama 3 in Aprilsetting a new performance standard in open source AI, then quickly followed up with Llama 3.1 in July with 405 billion parameters. Llama 3.1 builds were used to power Meta AI, the company’s built-in assistant across platforms like WhatsApp, Messenger, Instagram and Facebook, with the goal of becoming the most widely used AI assistant.

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

Meanwhile, Chinese models such as Alibaba Quinn Family-2.5 and New DeepSeek V2.5 and R1-Lite Preview Seemingly out of nowhere to top some benchmark charts, Nvidia itself has gone beyond providing graphics cards and software builds to launch its own powerful open source Model Nemotron-70B.

Nous Research, a small San Francisco company that aims to do more Customized and less restrictive AI models As open source, it also appeared for the first time several amazing new ideas.

And let’s not forget France Mistralwhich has rapidly expanded its open-source and proprietary AI offerings.

3. Google’s Gemini series has become a serious contender for the best available

In this year’s comeback story, Google’s Gemini series of AI models that were once ridiculed for their weird image generations and criticized for being too “woke” have returned, with new, more powerful versions that now top third-party performance benchmark charts and are increasingly appealing to developers and enterprises.

Google provided Gemini 2.0 Flasha multimedia AI model that supports video streaming analysis and can see and direct what you are doing on your screen, and follow up with Gemini 2.0 Flash Thinking Which competes with OpenAI’s o1 and o3 inference models.

4. AI has invaded the establishment

As the year went on, “agent” AI went from being a world of hype to a veritable series of product announcements and major initiatives by major enterprise software vendors. Take for example:

Salesforce’s Agentforce 2.0: Salesforce has unveiled Agentforce 2.0 A few days ago, advanced AI agent software was launched to enhance ideation, integration and personalization features across CRM and sales offerings, as well as Recessionand significantly improve enterprise productivity tools.

Goal SAP: SAP has converted its Joule chatbot to The AI ​​agent is powered by open source large language models (LLMs).), leading to innovation and efficiency in enterprise settings.

Google Astra Project: As part of its Gemini 2.0 initiative, Google launched Project Astra, an AI-powered assistant designed to provide real-time contextual responses by leveraging Google’s suite of services, with the aim of enhancing user productivity and decision-making.

My big prediction for 2025: AI-generated content will take precedence

Building on these developments, 2025 is expected to see AI-generated content spread across business and consumer industries, with everyone from OpenAI to Meta, Google, Microsoft, Apple, and even… Elon Musk’s xAI device now has AI-powered image generators Built into their offerings.

This expansion will simplify content creation, enhance personalization, and increase efficiency across sectors.

Additionally, we anticipate large-scale initial deployments of large language models (LLMs) and AI-powered generative robots in both commercial and consumer environments, which will revolutionize automation and human-robot interactions.

That’s all for the final #AIBeat newsletter of 2024. Thank you for reading, writing, subscribing, sharing, commenting and for being here with us. We look forward to sharing more and hearing more from all of you in 2025.

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