Let's not kill AI Innovation with SB 1047

The proposed California legislation, SB 1047, has sparked significant concern within Silicon Valley's tech community.?

Many fear it could significantly impede AI innovation in California and potentially lead to an exodus of companies from the state.?

Governor @GavinNewsom has been very clear that he wants California to lead the way with AI and will not yield ground. We should hit the kill switch with this particular bill.

Here's my analysis of the bill - the good, bad and the ugly:

It has passed the Senate and will be up for a vote in the assembly. Call your Assemblymember https://findyourrep.legislature.ca.gov/ and ask them to NOT support it.

Send an email to Governor Gavin Newsom https://www.gov.ca.gov/contact/ requesting him to veto the bill if it ends up at his desk.

Talking points are below:

Innovation impact: The bill's requirements will slow down the pace of AI development in California.

SB 1047, the Safe and Secure Innovation for Frontier Artificial Intelligence Models Act, could potentially hinder AI innovation in California. If similar legislation is adopted in other states, it may impact America's competitive position in AI development. The new Board of Frontier Models within the Government Operations Agency in California? would be unprecedented and there is a push in Washington already to consider various options.

Economic consequences: There's potential for reduced tax revenue and job losses if companies relocate. Can California be able to withstand this given our burgeoning deficit today?

Competitive disadvantage: California and potentially the U.S. could lose ground to other regions with less restrictive regulations.?

California has long been at the forefront of technological innovation. However, the potential consequences of this new legislation raise important questions about the state's future leadership in this arena.

If SB 1047 is enacted and leads to the anticipated negative impacts on the AI industry, how will Sacramento address these outcomes? Will policymakers take responsibility for any economic or innovative setbacks that may result from this legislation?

Implementation challenges: The bill's provisions may present practical difficulties for companies to comply with. The exact definition and scope of AI Covered models is confusing, AI regulation based on fixed FLOP count or $100 million to train? is a regulation that can go out of scope quickly.?Arbitrary compute thresholds are not very well thought out - as we will outgrow the current state very quickly!

Section 230 applies to the internet FDA regulates drug development We need something in between for AI

Unintended consequences: Some aspects of the bill could have unforeseen effects on the AI industry and its development. It's important to consider both the intended benefits of such regulation and its potential impacts on innovation and economic competitiveness

Balancing act: While the bill aims to address important concerns, it may overcorrect and stifle progress. We are in the very early stages of AI development and creating roadblocks will not bode well.

Industry response: Many tech leaders and companies have expressed strong reservations about the bill. The bill appears to have some areas that may benefit from further refinement. The bill’s approach to various issues, including potential existential risks, may have unintended consequences or gaps.

SB 1047's proposal to impose civil and criminal penalties for AI-related issues would set a unprecedented precedent in software regulation. This approach raises questions about proportionality and consistency in tech legislation.

Consider the past: We didn't pursue legal action against Microsoft when their operating systems crashed or required manual restarts. Similarly, the misuse of aircraft in terrorist attacks didn't result in lawsuits against plane manufacturers.

This comparison highlights the unique treatment proposed for AI under SB 1047. It prompts us to consider:

1. Is this level of legal liability appropriate for software development?
2. How does this compare to how we've handled technological risks in other sectors?
3. Could this approach deter innovation and risk-taking in AI development?
4. Are there more effective ways to ensure AI safety without resorting to criminal penalties?

These questions underscore the need for a balanced approach that promotes both innovation and responsible development in the AI sector.

This will undermine the future of open source AI development

Startups will not be able to handle the legal bills and thus unable to compete with the big boys

The system would be stacked against startups and create an unfair system.

It is prudent for Sacramento to bet on the future of AI and focus should be

A increase funding for AI innovation

B Address the job loss that is coming

C Address deep fakes and use by malicious actors

D Develop an AI education curriculum for K-12

E Mitigate any competitive IP thefts and ensure California’s success

The onus should be upon agencies discovering malicious users of AI and not putting the burden on the startups. Just like highway patrol tickets, people driving at over the speed limit.?

Please take action today.

Call your Assemblymember and also ask Governor Newsom to veto this bill if it ends up at his desk

That’s it for now. Stay tuned for updates below - so bookmark this.

If you enjoyed this thread:

1. Share this post and retweet this?

https://x.com/rishikumar1/status/1808164048711962795

2. Don’t forget to follow @rishikumar1 on X

For tech updates, connect with me on LinkedIn