The On Ai Should Not Make Things Worse Essay

1697 Words7 Pages
As time progresses companies that work on AI have modified the original laws to better suit and apply to modern and developing technology. For example Google’s developers came up with five “practical research problems” that developers should focus on when attempting to create AI. The reasons are as follows: 1: “Avoid negative side effects.” - AI should not make things worse. 2: “Avoid reward hacking.” - AI should not cheat to achieve its goals. 3: “Scalable Oversight.” - AI should look to humans for decision making. 4: “Safe Exploration. “ - Allow AI to learn at its own pace 5: “Robustness to Distributional Shift.” - AI should realize it’s exactly that, artificial.[2] Another example of “laws” that are relevant to AI are the Rules and Principles that were created by Microsoft’s Satya Nadella. His rules are specifically designed so that only AI should obey. However his rules apply more to the social impact that AI should have on the general populous. His six principles state: 1: “AI must be designed to help humanity.” - AI should do things that humans can’t while still respecting the autonomy that humans have. 2: “AI must be transparent.” - How AI works should never be a secret from the audience it affects 3: “AI must maximize
Open Document