Thank you for writing this. As an independent solopreneur in the AI space and software engineer, this all feels dreadful. The EU AI act, as well as other EU 'digital' regulation seems poised to destroy the European software industry, if not by intentet then by de facto outcome. Many of the EU software providers operate in a niche market, often bootstrapped by their owners. This industry is not ready to deal with the barrage of new EU regulationa, aka "the blue wall" - neither in terms of required legal expertise nor in terms of capital.
I think there is some sensibility in the heavy compliance requirements for those AI systems that are classified as "high-risk" under the AI Act. Do you want kids to be taught by an AI tutor? If they are, I think it's in order to do some safety testing and comply with a few documentation requirements. Sure, the AI Act wil have economic costs for AI companies but as a whole society will benefit from AI that is responsibly built and keeping humans accountable for any dangers and risks their AI systems may present.
I see the problem with EU bureaucracy and tendency for overregulation, but possible negative externalities, with systems so powerful, new and not yet fully understood, should not be underestimated. Why rush so fast? Many leading AI scholars and researchers are pointing out those risks not without merit.
Good article! It could do with a section on the extreme risk models though - AGI development should only happen veeeeery carefully and requires clear & international regulation. But obviously, other models should not be lumped into the same category
Maybe Brexit wasn’t so bad after all—at least the UK isn’t tied to the EU’s strict AI Act. Glad to see a more flexible approach that could actually support innovation!
Thank you for writing this. As an independent solopreneur in the AI space and software engineer, this all feels dreadful. The EU AI act, as well as other EU 'digital' regulation seems poised to destroy the European software industry, if not by intentet then by de facto outcome. Many of the EU software providers operate in a niche market, often bootstrapped by their owners. This industry is not ready to deal with the barrage of new EU regulationa, aka "the blue wall" - neither in terms of required legal expertise nor in terms of capital.
I think there is some sensibility in the heavy compliance requirements for those AI systems that are classified as "high-risk" under the AI Act. Do you want kids to be taught by an AI tutor? If they are, I think it's in order to do some safety testing and comply with a few documentation requirements. Sure, the AI Act wil have economic costs for AI companies but as a whole society will benefit from AI that is responsibly built and keeping humans accountable for any dangers and risks their AI systems may present.
I see the problem with EU bureaucracy and tendency for overregulation, but possible negative externalities, with systems so powerful, new and not yet fully understood, should not be underestimated. Why rush so fast? Many leading AI scholars and researchers are pointing out those risks not without merit.
As I just found this More in Common report, almost half of US surveyees believe gov should do more to regulate AI (vs 17% saying it is regulating too much) - https://www.canva.com/design/DAGRVonfEGk/W6C6mAYwwJmRhS2dqjjfVg/view#52
Good article! It could do with a section on the extreme risk models though - AGI development should only happen veeeeery carefully and requires clear & international regulation. But obviously, other models should not be lumped into the same category
Maybe Brexit wasn’t so bad after all—at least the UK isn’t tied to the EU’s strict AI Act. Glad to see a more flexible approach that could actually support innovation!