I've been closely following the developments surrounding California's SB1047, which has recently passed the assembly. Andrew Ng's recent Time article on this legislation highlights critical issues that I believe warrant serious attention.
As a professional deeply involved in the AI industry, I find myself compelled to speak out against SB1047. This bill, now awaiting the governor's decision, is fundamentally flawed. Its passage through the assembly reveals a concerning lack of understanding among lawmakers about AI technologies and their implications.
The issues with SB1047 are numerous and significant, as detailed in Ng's article. These flaws could potentially stifle innovation and create unintended consequences for the AI industry in California and beyond.
Given the bill's deep-seated problems, I believe Governor Gavin Newsom should veto SB1047. The appropriate path forward is to draft an entirely new bill, one that is crafted from the ground up with substantial input from AI experts who truly understand the nuances and complexities of this rapidly evolving field.
Effective AI regulation is crucial for our future, but it must be grounded in a comprehensive understanding of the technology and its potential impacts. We need legislation that addresses legitimate concerns while also fostering innovation - a balance that SB1047, in its current form, fails to achieve.
I encourage my colleagues in the AI community to engage in this important dialogue. It's crucial that we lend our voices and expertise to shape policies that will responsibly guide the development and deployment of AI technologies in California and beyond.
What are your thoughts on SB1047 and the future of AI regulation? How can we ensure that future legislation is better informed and more effective?
tbh really underestimated how much people around you shape you and affect your happiness. i was lucky to discover the kinds of people i enjoy spending time with – curious what happens when you hang around them full-time ;) @theresidency
@yacineMTB i believe portions of the matrix covered by a kernel on a given pass can be unfolded into vectors to transform a convolution into a matmul. look up fold/unfold in pytorch!