https://www.nature.com/articles/d41586-023-04043-w Nature reports that deep seek has discovered some results that exceed the ability of people to have created!
Hurrah, so now we can sit back and let the AIs take over the act of science?
Not so, not so at all. For these devices have no agency, they are just tools.
It would be akin to saying that because a new telescope has seen further we no longer need astronomers.
Without the ability to have agency and curiosity these tools are just that, complex extraordinary and as yet not fully harnessed tools.
It turns out that we don’t quite know the limits of these new tools, but the domains of the world that they can be applied to at present will remain in the domain of symbol manipulation and data analysis. Not in the domain of observation of phenomena or the creation of new data, or indeed the undirected creation of new questions.
This things might change, and there may be simple genetic loops that we can place these tools into that will emulate some of the behaviour, but it will take someone with curiosity, funding, time, and skill, to kick off activities like that.
It might be more accurate if our media reported findings from these tools as if they were findings enabled by new equipment, rather than findings aneled by new intelligences.
Hurrah, so now we can sit back and let the AIs take over the act of science?
Not so, not so at all. For these devices have no agency, they are just tools.
It would be akin to saying that because a new telescope has seen further we no longer need astronomers.
Without the ability to have agency and curiosity these tools are just that, complex extraordinary and as yet not fully harnessed tools.
It turns out that we don’t quite know the limits of these new tools, but the domains of the world that they can be applied to at present will remain in the domain of symbol manipulation and data analysis. Not in the domain of observation of phenomena or the creation of new data, or indeed the undirected creation of new questions.
This things might change, and there may be simple genetic loops that we can place these tools into that will emulate some of the behaviour, but it will take someone with curiosity, funding, time, and skill, to kick off activities like that.
It might be more accurate if our media reported findings from these tools as if they were findings enabled by new equipment, rather than findings aneled by new intelligences.