I wonder for some time now why only very few people mention the bad user interface of chat bots and chat UIs in general. There simply are no affordances, as Don Norman would have said!
I think some of the opportunities that AI offers are in the user experience layer—it presents the potential of an interface that can adapt its paradigm to the user's conceptual model rather than, at present, where users must adapt their conceptual model to that of the application.
Text chat isn't its final form, but I don't know where it will evolve or whether it will become more commonplace than the GUI systems we use today. Will adaptive GUIs be a thing? Interacting with humanoids? Or something else?
That's exactly my point. We don't know yet what new interaction models will be enabled through this new technology. But I think we will have big differentiators there.
I see some early different approaches in Dev-Tools, ranging from a focus on code to the output (e.g., Cursor vs. Lovable).
I would love to see more experimentation happening, the big model builders are currently very boring and apparently not thinking that much about how to fit these tools to actual workflows and opportunities.
Definitely, I think that's by design - they want to be utilities - the next power company who charge for every use and want us to experiment and discover the new paradigms on their behalf.
Yes, it's unclear what OpenAI, Anthropic, etc. want to be currently. My theory is that they want to be the LLM compute platform, delivering their services through APIs. But now they get more pressure from the quick commodisation of models. It doesn't look like pure models are a great business case.
Hiring a CPO from Instagram/Twitter sounds like they see potential in some form of B2C model, but we will see. How to make money from creating general models is still an open question.
Absolutely i's difficult to build a defensible competitive advantage with commodity AI. The massive funding rounds for OpenAI, Anthropic etc. will have them searching for all manner of revenues.
Sort of like how large funding rounds of ride share and messaging apps appeared to create the super app phenomenon. It's going to be a wild ride. I wouldn't be surprised if they bought their hardware & data center suppliers either.
It looks like Windsurf is about to be acquired too.
Great point about that hire signalling a greater consumer focus - the financial obligations mean they need to take finding killer apps into their own hands...
I wonder for some time now why only very few people mention the bad user interface of chat bots and chat UIs in general. There simply are no affordances, as Don Norman would have said!
I think some of the opportunities that AI offers are in the user experience layer—it presents the potential of an interface that can adapt its paradigm to the user's conceptual model rather than, at present, where users must adapt their conceptual model to that of the application.
Text chat isn't its final form, but I don't know where it will evolve or whether it will become more commonplace than the GUI systems we use today. Will adaptive GUIs be a thing? Interacting with humanoids? Or something else?
That's exactly my point. We don't know yet what new interaction models will be enabled through this new technology. But I think we will have big differentiators there.
I see some early different approaches in Dev-Tools, ranging from a focus on code to the output (e.g., Cursor vs. Lovable).
I would love to see more experimentation happening, the big model builders are currently very boring and apparently not thinking that much about how to fit these tools to actual workflows and opportunities.
Definitely, I think that's by design - they want to be utilities - the next power company who charge for every use and want us to experiment and discover the new paradigms on their behalf.
Yes, it's unclear what OpenAI, Anthropic, etc. want to be currently. My theory is that they want to be the LLM compute platform, delivering their services through APIs. But now they get more pressure from the quick commodisation of models. It doesn't look like pure models are a great business case.
Hiring a CPO from Instagram/Twitter sounds like they see potential in some form of B2C model, but we will see. How to make money from creating general models is still an open question.
Absolutely i's difficult to build a defensible competitive advantage with commodity AI. The massive funding rounds for OpenAI, Anthropic etc. will have them searching for all manner of revenues.
Sort of like how large funding rounds of ride share and messaging apps appeared to create the super app phenomenon. It's going to be a wild ride. I wouldn't be surprised if they bought their hardware & data center suppliers either.
It looks like Windsurf is about to be acquired too.
Great point about that hire signalling a greater consumer focus - the financial obligations mean they need to take finding killer apps into their own hands...