Right now bots are partly/mostly DIY for capabilities, but improvements will be made and then distributed. Future versions will come with lots of useful and general capabilities onboard. Model development will increasingly be directed at making bots work better. And improvements in memory systems will be valuable and so big teams will work on them then sell the better systems.
There will still be individual contributions and choices like having the bot work on itself and learn new capabilities.
Of course the obvious outcome is that very shortly, bots won’t need human direction, and will work long-term on projects humans set in motion, or misunderstandings/distortions of human instructions.
Right now bots are partly/mostly DIY for capabilities, but improvements will be made and then distributed. Future versions will come with lots of useful and general capabilities onboard. Model development will increasingly be directed at making bots work better. And improvements in memory systems will be valuable and so big teams will work on them then sell the better systems.
There will still be individual contributions and choices like having the bot work on itself and learn new capabilities.
Of course the obvious outcome is that very shortly, bots won’t need human direction, and will work long-term on projects humans set in motion, or misunderstandings/distortions of human instructions.