What did Churchill say? “Success is stumbling from failure to failure with no loss of enthusiasm.” So it goes in product development. Zuck was hard at work defining the Metaverse, even going so far as to rename the company. Wow! So far though, that’s been kind of “meh!” But let’s not call him shy. Yes, Threads is an interesting play garnering “fastest growing app in history.” But, let’s face it, Threads is a copy of Twitter with automagically imported users. And so far, only what, 5% of users have migrated over. Will the chore of doubling your social media burden prove too much? We shall see.
What’s more interesting for Meta though is their approach to AI where they’re about to release an updated version of their AI model LLaMA. Unlike its competitors, Meta’s LLMs are open-source, meaning the details of the new model become publicly available. This is in contrast to models like OpenAI’s GPT-4, which is a “black box” where the data and code used to build the model are a secret. Meta’s move towards open-source AI is their attempt to change the competitive landscape of AI, allowing companies of all sizes to improve it, to build applications on it, which theoretically could help Meta catch up with its rivals.
While LLaMA is currently free and open-source, Meta’s been exploring the possibility of charging enterprise customers for the ability to fine-tune the model using their own proprietary data. Yes! And guess what, this actually does tie back to the larger goal of building a digital world known as a metaverse. Despite the constant slings and arrows, Zuck has not abandoned this goal.
Open-source models do have advantages, a higher uptake by users who provide more data for the AI to process, the ability for researchers and developers to spot and fix bugs. However, however, open-source AI also comes with risks, such as potential misuse by bad actors.
The bottom line is, competition demands innovation. And innovators need the chutzpah to launch products that “might not work.” Innovators don’t assume perfectly planned and anticipated outcomes. For Meta, “this might just work.”
Related: Reading, Actually: The Lost Art of Reading in 21st Century Culture