“Every 18 months, the minimum IQ required to destroy the world drops by one point,” says Eliezer Yudkowsky, an AI theorist and co-founder of the Berkeley-based Institute for Machine Intelligence. put forward with an obvious improvisation of Moore’s Law. The degree of existential risk posed by AI is OpenAI ChatGPT, it may seem overstated for now, but policymakers across jurisdictions are stepping up regulatory scrutiny of generative AI tools. Reported concerns fall into three broad categories: privacy, system bias, and intellectual property violations.
Policy responses are also different, with the European Union predictably tougher by proposing to introduce new AI legislation that segregates artificial intelligence by use-case scenario, broadly based on degree of invasiveness and risk. I’m taking a stance. The UK is on the other end of the spectrum, with a decidedly ‘light touch’ approach aimed at promoting rather than stifling innovation in this nascent sector. The U.S. approach falls somewhere in between, with Washington poised to define an AI regulatory rulebook by starting public consultations on how to regulate artificial intelligence tools earlier this month. This is ostensibly based on a move by the White House Office of Science and Technology Policy last October to release a blueprint for his AI Bill of Rights. China has also announced its own set of measures to regulate AI.
India has said it is not considering legislation to regulate the artificial intelligence sector, and Ashwini Vaishnaw, the coalition’s IT minister, said AI “had ethical concerns and risks associated with it”. It has proven to be an enabler for digital and innovation ecosystems.
“NITI Aayog has published a series of papers on the subject of Responsible AI for All. However, the government is not considering enacting laws or regulating the growth of artificial intelligence in the country. ‘, he said in a written response to the House of Commons on the Budget.
american approach
On April 11, the U.S. Department of Commerce took the most decisive step to address regulatory uncertainty in this area, creating rules and laws to ensure AI systems perform as advertised. We asked the public to comment on the methods. Authorities flag potential floating audit systems to assess whether AI systems contain harmful biases or distort communications to spread misinformation or disinformation. I got
According to Alan Davidson, assistant secretary of commerce at the U.S. Department of Commerce, much like an auditor checks the accuracy of a business report, ensuring AI systems work without adverse consequences , may require new assessments and protocols. The catalyst for all this policy action in the US appears to be his October 2022 move by the White House Office of Science and Technology Policy (OSTP). A non-binding roadmap for the responsible use of AI. The 76-page document details five core principles for governing the effective development of AI systems, with particular attention to the unintended consequences of civil and human rights violations. That’s right.
A safe and effective system: Protect users from insecure or ineffective systems
Algorithmic Discrimination Protection: Users do not have to face algorithmic discrimination
Data privacy: Users are protected against data misuse with built-in protections and powers of attorney over how their data is used
Notice and Explanation: Users are aware that automated systems are being used and understand how and why automated systems contribute to the outcomes that affect them.
Alternative option: Users can opt out and access personnel who can quickly review and resolve any issues they encounter.
The Blueprint expressly states that it “helps guide the design, use, and deployment of automated systems to protect the American public,” and its principles are non-regulatory and non-binding. There is still no advertised “blueprint” and no enforceable “Bill of Rights” with legal protections.
The document includes several examples of AI use cases that the White House OSTP has deemed “problematic” and “have meaningful implications for the rights, opportunities, or access to critical information of American citizens.” It should only be applied to automated systems where it could be impacted.” Resources or services, generally excluding many industrial and/or operational applications of AI. According to a World Economic Forum outline, the blueprint expands on examples of using AI in lending, human resources, surveillance and other areas, and will be included in the proposed EU AI law’s “high-risk” use case framework. You can also find a match for document.
But analysts point to gaps. Nicol Turner Lee and Jack Malamud of Brookings suggest that identifying and mitigating the intentional and unintended consequential risks of AI has been widely known for quite some time, but how such complaints can be reprimanded is a blueprint. “Furthermore, questions remain as to whether a non-binding document will prompt the necessary congressional action to govern this unregulated space.” they said in a December paper entitled Opportunities and Blind Spots in the White House AI Bill of Rights Blueprint.
The regulatory debate has gained momentum in the wake of developments surrounding the soft launch of ChatGPT, OpenAI’s chatbot based in San Francisco, which is estimated to have more than 100 million users, with Google bard Bard I am developing a chatbot. The Chinese company follows Baidu’s Ernie Bot and Alibaba announcing plans to release a bot for internal use.
Pause AI development
Tech leaders Elon Musk, Steve Wozniak (Apple co-founder), and more than 15,000 others have responded by calling for a six-month moratorium on AI development, and the lab will ban systems no one builds. It says it is in an “uncontrollable race” to develop. You have full control. They also said labs and independent experts should work together to implement a set of shared safety protocols. Yudkowski is also one of those calling for a global pause in AI development. But the call further divided opinion.
“Demands to suspend work on models more advanced than GPT-4: This is regressive when we are cracking down on technology that could prove harmful to society. Anything left unattended and unregulated can prove harmful.Think about monetization, regulation, prudent use of LLMS and related technologies instead of asking for a moratorium There is a need,” Anuj Kapoor, assistant professor of quantitative marketing at IIM Ahmedabad, told The Indian Express.
We’ve seen a flurry of policy activity in the US, but given repeated calls for the US Congress to pass laws that limit the power of big tech, how much progress is likely to be made in Washington on this issue? I’m not very optimistic about Given the political divisions among lawmakers, the attempt has made little headway.
The EU seems cautiously wrong given Italy has set the stage by emerging as the first major Western country to ban ChatGPT over privacy concerns. The 27-member bloc pioneered in 2018 by initiating measures to regulate AI. The EU AI law, scheduled for 2024, is therefore a long-awaited document.
China has a regulatory regime in place for the use of AI. Earlier this month, the country’s federal internet regulator released a draft of 20 items regulating generative AI services, including obligations to ensure accuracy and privacy, prevent discrimination, and guarantee intellectual property rights. bottom.
Open for public feedback and likely to come into force later this year, the draft establishes a mechanism for AI providers to clearly label AI-generated content, handles user complaints, and makes it public. It also requires a security assessment before According to the draft cited by Forbes, AI-generated content must “reflect core socialist values,” including the overthrow of state power that could lead to the overthrow of China’s socialist system. should not be
Incidentally, the Chinese regulations were published the same morning the US Department of Commerce asked for comment on AI accountability measures.
