Eric Schmidt, the leader of the U.S. government initiative to integrate AI into national security, warned that the EU’s AI transparency requirements would be ‘very harmful to Europe.’ Schmidt spoke at POLITICO’s AI Summit and criticized the provisions of the European Union’s AI Bill.
In the EU AI bill, the requirement is that the AI algorithms have to be transparent, which seems reasonable. Schmidt thinks that should the proposal go through in its current form, it will be a big setback for Europe. He proposes that modifications be made.
Why modify the bill?
Schmidt is the chair of the National Security Commission on Artificial Intelligence (NSCAI) and worked as the CEO of Google before that. The EU’s proposal requires the AI to ‘explain itself.’
However, AI algorithms cannot always fully explain how they arrived at some decisions.
Schmidt sees this as crippling. However, people concerned about the terrible power of AI are wary of letting these systems run autonomously without explaining why they arrive at decisions. On the surface level, it seems better to know why something is happening than to just blindly follow complex suggestions.
The former Google exec distanced himself from the government body he chairs, saying that the remarks were his own and not those of the commission. It is suspicious that someone in his position wants AI to be implemented in a non-transparent way.
For a long time now, laws that reduce or minimize responsibility for tech companies have been the mainstay and have allowed them to commit a laundry list of crimes, most notably the invasion of personal privacy.
Schmidt is not the only critic of EU policy. He is part of the US establishment that wants less stringent rules, calling the EU’s ‘regulation first stance a mistake, saying that they should have started by being the US’s ‘innovation partner’ to compete with China.