OpenAI CEO Sam Altman testified before the Senate Judiciary Subcommittee on Privacy, Technology, and Law last month. Wynn McNamee — Getty Images
Open source code has revolutionized the world of technology, but recent developments such as the rise of foundational models, accelerating investment in artificial intelligence, and the growing geopolitical AI arms race have pushed the open source community to open source. We are forced to face the ethical issues surrounding the code. .
Potential intellectual property violations, perpetuation of bias and discrimination, privacy and security risks, power relations within communities, governance issues, and environmental impacts are all ethical issues that need to be addressed.
These issues have started discussions about whether moving from the open source movement to ethical sources could be the solution. Many developers have advocated licenses (much like the Hippocratic license) that impose ethical restrictions on the use of open source code. Some point to the potential role of government regulators as a solution.
There are many unknowns when it comes to machine learning models. Developers of these models will be faced with a decision to open source or not. But it is no more possible for developers to predict all possible use cases for machine learning than it is for mathematicians at Bletchley Park to predict all potential use cases for computers. Most developers realize that when they open up their code, they lose control over how it is used and by whom.
Licensing is gaining more and more attention as a solution to this problem. But restricting the use of open source code with additional licenses not only contradicts the core principle of the open source community that code should be accessible to everyone, but also undermines the collaborative working environment that is the foundation of open source. There is also the possibility of loss. The community’s ability to accelerate technology development.
Also, there are many questions about whether ethical licensing actually reduces the risk of code being used for illicit purposes. Many countries have already enacted human rights laws, and individuals and organizations that violate human rights laws should be prosecuted accordingly, regardless of the methods or technology used to carry out the abuses. If such laws fail to deter violators, license agreements are unlikely to influence offender behavior.
AI systems are complex, and enforcing their ethical use is cumbersome. Rapid advances in AI technology make it difficult to keep up with developments and potential ethical implications. Additionally, the lack of transparency and accountability in AI systems can make it difficult to hold organizations accountable for ethical violations. Addressing these challenges will require continued collaboration, dialogue, and commitment to ethical research and development to ensure that AI is used in ways that align with societal values and promote greater good. Requires investment.
The burden of ethical use should fall on those who use open source code to build AI products, not those who write the code. This is why government regulation is key to ensuring the ethical use of AI. Government regulation enforces strict definitions of ethical use and creates bureaucratic structures around the evaluation of AI systems. leads to
Just as governments regulate and scrutinize medical products before approving them for public consumption, governments can also ensure that AI passes certain tests before being released to the public. The responsibility should fall to governments with sufficient resources to explore these tools, not to developers who can focus on building more advanced AI.
Any company, organization, or individual should provide clear details about the nature and broader implications of their model, including the data used to train the model and the code used to develop it. Potentially harmful applications must be disclosed before being made available. This application approach is not too different from the application process used by many large academic conferences. This application process enforces a submission structure that requires applicants to address ethical issues and broader implications with transparency.
In general, evaluating models to a level where users can be confident that they are always behaving ethically and correctly remains a problem in the AI field. Even OpenAI is struggling to find a way to do it with ChatGPT. The good news is that with a lot of research being done to understand fairness, accountability and transparency in AI models, governments already have many tools to start building regulatory frameworks for AI. about it.
The European Union has released a proposal for a new legal framework on AI in 2021, aimed at ensuring that AI is developed and used in a manner consistent with EU values. The United States has established the National Artificial Intelligence Initiatives Office (NAIIO) to coordinate federal investment in AI research and development. Canada likewise funded interdisciplinary research on AI and established the Canadian Institute for Advanced Study (CIFAR) to develop ethical and technical standards for AI. Singapore has introduced the Model AI Governance Framework, which provides guidance on the responsible development and use of AI.
By adopting a multi-stakeholder approach and investing in ethical research and development, governments can ensure that AI is developed and used in ways that are consistent with societal values and promote greater good. increase.
We are moving into an era where decisions are made for and about individuals using algorithmic processes without human involvement. Individuals have the right to an explanation of how the AI system reached its decisions, and that explanation depends on having a transparent process.
The ethical source licensing movement is born of strong principles and positive intent, but its effectiveness is limited by factors such as lack of enforcement mechanisms, limited scope, lack of awareness, alternative licensing options, and lack of standardization. is. This approach requires ex post legal effort. Licensors must monitor all usage of open source code, which is not practical. Governments, on the other hand, have an opportunity to protect their citizens from the potential adverse effects of AI systems, and they can do so through aggressive regulation and enforcement.
Frederik Hvilshoj is a machine learning leader. encode.
The opinions expressed in commentary articles on Fortune.com are solely those of the authors and do not necessarily reflect the opinions or beliefs of the authors. luck.
