Google and OpenAI are two of the largest AI companies in the world, investing hundreds of millions of dollars towards developing cutting-edge technologies. However, the rapidly multiplying open source projects that are popping up may be their greatest threat yet. A document circulating within Google stated that both companies had little to no “moat” surrounding them, meaning they could easily be taken down by smaller, more agile competitors.
The AI world moves at a breakneck pace, dwarfing the usual speed of traditional software updates. This means that while OpenAI and Google have poured resources into deep learning models such as GPT-3, ChatGPT and GPT-4, the development of open source projects has far exceeded the scope of those releases. Soon after Meta leaked its LLaMA language model in March, developers had already adapted, forming new core components like instruction tuning, multiple modalities, and reinforcement learning. No matter the resources of OpenAI and Google, it was impossible for them to keep up with the speed of the open source projects.
What this demonstrates is that the “Wal-Mart” hulks of the AI world, like GTP-4, might not be the go-to solution in the long run. While GPT-4 is supposedly suited to answering questions on contract law, why should companies invest in such a bulky invention when they can get a smaller, more inexpensive solution? The concept of “smaller is better” has already proven itself before, with image editing projects: Photoshop has been completely replaced for most people.
Google has urged its competitors to embrace the open source movement, publishing, sharing and compromising. Sam Altman of OpenAI has previously noted that the problem is not solved by throwing more computing power at it. It’s clear that OpenAI and Google have been too reliant on the sheer size of their projects in the past, and now they must learn to adapt to the threats of the open source revolution. People have already begun to realize that these companies have chosen AI models to fit their own existing business model, and no longer necessarily the other way round. Ultimately it’s the small startups and individuals who will reap the rewards, as long as they remain in tune with the fast-paced development of the open source landscape.
