The article also highlights the political implications of AI, particularly in relation to AI regulation in the US and EU. Gilbert warns of the dangers of AI becoming a political project without rigorous scientific evidence to back up claims about its potential risks and benefits. He calls for greater accountability in AI research and development, arguing that without it, we risk neither good politics nor good science.
Key takeaways:
- Machine ethicist Thomas Krendl Gilbert argues that much of today's AI research is more akin to alchemy than science, due to its lack of scientific rigor and the secrecy surrounding its development.
- Gilbert suggests that the "magic" of AI comes from the interface, not the model itself, and that AI builders often treat the technology as if it has magical properties.
- He warns that the lack of understanding and transparency around AI mechanisms can lead to problematic intersections with politics, particularly in discussions around AI regulation.
- Gilbert calls for more accountability in AI development, arguing that without it, the field risks becoming neither good politics nor good science.