ED Antidote to AI hype
Transparency essential to regulate tech advancement
By Daniel Shin
AI is a powerful technology that can have significant impacts on various aspects of society, such as the economy, politics, security and ethics. Therefore, it is important to have a check-and-balance system as AI expands exponentially to ensure its innovative but safe and ethical use.
Some experts suggest that AI systems must be compartmentalized regarding their functions and domains, to prevent the concentration of power and influence in a few tech powerhouses. This could help balance the power dynamics of AI across different sectors and stakeholders.
In order to get there, we must design AI systems with human oversight, accountability and technological transparency in mind. Technological transparency implies that AI systems must be auditable, comprehensible and easily understood by human beings at varying levels of comprehension and expertise as the European Commission rightly said.
However, is it fair for those who have poured money into their AI ventures to develop state-of-the-art technologies? The government could instead intervene and regulate business model transparency. Business model transparency means that human beings are knowingly informed of the intention of business owners or developers of AI systems. That connects directly to commercial viability.
Now, as many companies race to the peak, severe competition is on the horizon. There will be many companies that will come and go in the AI industry, and it will not be easy for them to survive. However, competition does not justify hindsight. Technology implementers should, therefore, test and validate AI systems in different scenarios and environments to ensure their reliability, robustness and predictability. They should direct their efforts to detect and self-correct any errors, biases or harms.
Innovation hates regulation. Regulation is often blamed for inhibiting innovation. But regulation and innovation do not have to be enemies. Regulating AI systems and governing underlying data with clear and consistent rules, standards and principles is much needed and must be aligned with human values and interests.
AI regulation is a complex and controversial topic that involves many stakeholders for both opportunities and risks. There are different arguments on regulating AI, depending on the perspective, goals and values of each party involved. Innovators are worried about over-regulation or under-regulation of AI systems before they even bloom. Regulation could stifle innovation or create unintended consequences. Some forerunners however may already want some level of regulations on newcomers to create barriers to entry.
The AI competitive landscape is complex. Furthermore, AI is now a global affair. Artificial intelligence is taking the world by storm. To address the global and cross-border implications of AI systems and development challenges, we must ensure coherent coordination among different jurisdictions. Many states acknowledge their limits and the challenges of regulating AI systems. Therefore, they must collaborate to set a new form of governance, oversight or law enforcement.
AI systems are autonomous to the extent that they can now make individual decisions without direct human interference. But business decisions are not purely autonomous. Business decisions are subject to moral constraints especially when the public interest evokes human values. To respect the autonomy, diversity and creativity of AI systems, users should be a catalyst to experiment, learn and improve AI systems without excessive constraints or interference.
Voluntary regulation of AI systems sounds like a dream. Leveraging the self-regulatory measures will, though, boost the longevity of innovation and creativity and reinstate publicity in the already-hyped AI industry. So far, the private sector, civil society and other nongovernmental actors have been flexible, adaptive and responsive to the fast-changing nature of AI. Therefore, AI leaders shouldn't be an enemy to them.
To foster the innovation, development and adoption of AI systems to a much greater degree than first imagined, all stakeholders should cooperate to create a level playing field for competition, collaboration and trust among different actors in the AI ecosystem. The basic principles should be to protect the rights, interests and values of humans. AI systems must be ethical, transparent, fair and respectful of human dignity and autonomy.
We need a clear and consistent definition of AI that covers its various forms, functions and applications. We need a mechanism for cooperation and coordination among different regulators, policymakers, developers, users and other stakeholders in the AI ecosystem, both within and across jurisdictions and regions.
The market needs an antidote to the hype of the AI saga. There is no definitive answer to what is really needed to regulate AI, as different stakeholders may have different views and preferences on how to approach this issue. But, we should keep pondering and come up with pragmatic solutions for that. AI, a transformative technological advancement, is supposed to change everything. But we shouldn't be the target.
Daniel Shin is a venture capitalist and senior luxury fashion executive, overseeing corporate development at MCM, a German luxury brand. He also teaches at Korea University.