Open source is critical to the future of artificial intelligence – here’s why

With all the media attention on AI mega-brands like OpenAI, Metaand anthropic, easy to miss huge ecosystem of artificial intelligence it rarely makes the news. I’m talking about the open source movement, which is busily creating a universe of artificial intelligence for those of us with thin wallets who still want to use those cool new tools.

Individuals and small companies driving this idea revolution in software not interested in making headlines; they are more interested in creating communities and celebrating the artificial intelligence ecosystem.

What is open source?

Open source is a software “movement” (they hate that term) that encourages the creation of free software. This means that the software is free to use, modify, and share with others. It’s all about open collaboration so everyone can contribute.

There are no restrictions on who can use it; The goal is to make the software as accessible as possible. This encourages people to work together and continue to improve products.

This concept originated in the early 1950s and continues to produce important results throughout the world. The Internet and World Wide Web would not be possible today without the open source software tools and standard protocols that run web servers and communications systems.

The role of stable diffusion in artificial intelligence

Stable diffusion

(Image credit: StabiltyAI)

The clearest example of this principle in action in the world of artificial intelligence is Stable Diffusion The universe of image generation. Since its launch in 2022, this text-to-image technology has spawned a huge army of developers creating tools and image services for the general public.

You can catch a lot of this on HugFace a website that hosts over a million open source models and demos aimed at improving the world of artificial intelligence as a whole.

Many of the great images you see on the market are the work of these people, and they are all released under the banner of free and open source software.

But it’s not just images and videos. There are also a sea of open-source artificial intelligence utilities that power applications far from the headlines. For example, Ollama a platform that brings a growing percentage of AI to users’ computers.

This downloadable technology allows you to run small artificial intelligence models on your computer, even if you have only modest settings. It’s great for hobbyists, privacy-conscious people, and people who resent the support of mega-cloud services like Microsoft and Google.

Creating Local AI

Before you start assuming it’s a nerd town, there’s a very good chance that if AI ever reaches your smartwatch or other small devices in your life, the groundwork will be laid by these kinds of home experiments.

That, and an army of open source artificial intelligence researchers at universities and colleges around the world. Never underestimate the power of a crowd.

However, not all open source work is done at the kitchen table level. Larger players such as Meta have put great effort into releasing so-called “open” large language models. In this case, they are only partially open, and there are growing calls to change the definition of LlaMa models from “open source” to “open weights.”

This is a technical but important difference to help maintain the purity of the open source method. Opening should mean that nothing is locked, which is not currently the case with LlaMa models.

Sometimes what’s open becomes closed

Hugging Face

(Image credit: Hugging Face/Nigel Powell)

At this point I’m going to add a big ugly caveat. There has been an unfortunate trend lately where startups are loudly announcing the launch of their new “open source” gold product, which is really just a marketing mechanism to attract public attention.

Once their numbers grow, they ignore the source code and release a full-fledged commercial paid service to bait and switch the sucker enthusiasts.

The most striking example, of course, is OpenAI. The company launched with great fanfare as a maker of open-source artificial intelligence models, attracting press attention, investment, and the brightest minds in the business. Then he quietly switched to profit-based enterprise last month. Pretty mean, Mr. Altman.

Open source will win in the end

Genmo Mochi-1

(Image credit: Genmo Mochi-1/AI generated)

Despite this and other similar moves, open-source artificial intelligence software continues to rock the Casbah as more people join the gold rush. It may not be money that drives these numbers, but reputation is also being built every hour in this hot tech environment.

Most names will never reach the public consciousness, such as TensorFlow, the pick and shovel of the AI modeling industry. But other names, such as the French models Mistral AI and BLOOM, are destined to gain public prestige very soon.

The true benefit of open source is the collaboration and improvements made by the community. One example of this is the discovery of the Llama metagame. This allowed languages to be preserved in countries that could never afford to create their own models.

We saw a shining example of the value of open source with Mochi-1 Genmo AI video model earlier this week. This is the most modern solution, comparable to such leaders as Runway and Luma Labs Dream Machine. At launch it will only run on a network of 4 expensive H100 GPUs, but within a few hours the community has created a customized version to run on a good gaming PC.