One question about the economies of scale here- what about the impact of loyalty and ease of switching. One reason SoMe platforms benefit from platform plays is that once you're on them- there are huge costs to switching. This is not something platforms like Uber and/or LLMs can replicate yet- which means they're much more susceptible to price wars.
Good callout. I agree that switching costs are an important factor in platform dynamics. However, economies of scale also play a crucial role here—more users mean more data, which helps improve models and user experiences. LLM providers can leverage this data to personalize interactions, like remembering user preferences or curating a specific writing or conversation style.
Memory is the key element in switching costs, which is likely why OpenAI has recently introduced features like memory and web search. These not only enhance user experience but also make switching providers less appealing.
Memory works ok-ish. It still often (a) misses information it should remember and (b) remembers irrelevant stuff. Both can be solved as GPT becomes better at tool use so hopefully a matter of time.
I was amazed when Meta launched their AI assistant and leveraged their massive existing platforms to instantly achieve what would take most companies years to build - an enormous user base for their AI by seamlessly integrating it into Instagram, WhatsApp, and other popular platforms where billions are already engaged.
One question about the economies of scale here- what about the impact of loyalty and ease of switching. One reason SoMe platforms benefit from platform plays is that once you're on them- there are huge costs to switching. This is not something platforms like Uber and/or LLMs can replicate yet- which means they're much more susceptible to price wars.
Good callout. I agree that switching costs are an important factor in platform dynamics. However, economies of scale also play a crucial role here—more users mean more data, which helps improve models and user experiences. LLM providers can leverage this data to personalize interactions, like remembering user preferences or curating a specific writing or conversation style.
Memory is the key element in switching costs, which is likely why OpenAI has recently introduced features like memory and web search. These not only enhance user experience but also make switching providers less appealing.
how good has it been for you. Personally not too impressed by it, but that could just be me
Memory works ok-ish. It still often (a) misses information it should remember and (b) remembers irrelevant stuff. Both can be solved as GPT becomes better at tool use so hopefully a matter of time.
I was amazed when Meta launched their AI assistant and leveraged their massive existing platforms to instantly achieve what would take most companies years to build - an enormous user base for their AI by seamlessly integrating it into Instagram, WhatsApp, and other popular platforms where billions are already engaged.
I recent in ok jnyhnjnjnnnnnnnbnn nnbx a