The leaked memo from within Google titled “We have no moat” discusses the significant progress made in artificial intelligence (AI) and challenges the assumption that only a few deep-pocketed firms would dominate the field. The memo argues that researchers in the open-source community, using free online resources, are achieving comparable results to proprietary models by fine-tuning existing large language models (LLMs) using a technique called low-rank adaptation (LoRa). The memo highlights the explosion of activity in open-source AI since the leak of the llama model created by Meta (Facebook’s parent company). This newfound accessibility to AI has implications for the industry’s future, democratising access to AI technology and lowering the barrier to entry for training and experimentation.
The memo suggests that Google and OpenAI may no longer have a defensive “moat” against open-source competitors, as the advantages of open-source, including speed, collaboration, and lower costs, cannot be easily replicated. However, not everyone agrees with this thesis, as the memo acknowledges that proprietary software coexists with open-source software in other domains, such as the internet. Benchmarking AI systems is challenging, and the balance between open-source and proprietary AI may resemble the balance found in other software domains.
The implications of this shift are both positive and negative. On the positive side, it reduces the likelihood of monopolistic control of AI, makes AI more affordable, accelerates innovation, and improves transparency and safety by allowing easier analysis of AI system behaviour. However, easier access to AI also raises concerns about bad actors fine-tuning systems for nefarious purposes, such as generating disinformation. It also presents challenges for regulating AI due to its widespread availability.
The impact on Google and other tech companies’ competitive advantage in AI will become apparent in due course. Nevertheless, this memo is seen as another turning point in computing, similar to previous influential memos by Bill Gates and Jeff Bezos, as it highlights the changing dynamics and potential disruptions in the AI industry.
(Source: The Economist)