It was rather unexpected to witness 28 national governments at the forefront of artificial intelligence (AI) signing a document advocating for AI safety, considering the usual lack of consensus among governments. This development follows President Joe Biden’s Executive Order, which introduced new requirements for AI safety. Noteworthy figures in the field, such as Elon Musk and Sam Altman, have also acknowledged the significant risks associated with AI despite the hype surrounding it.
Ritwik Gupta, an AI researcher and student at UC Berkeley’s Center for Security in Politics, expressed cautious support for this initiative. He emphasized, “These agreements signify a crucial step towards the responsible governance of AI in society.” The Executive Order empowers federal authorities to monitor and address potential threats to national security.
However, the question remains: are these measures adequate to ensure the health of AI? I would argue that they are not.
There are two primary reasons for this assertion. Firstly, these agreements primarily target government officials and AI developers, neglecting the corporate executives who intend to utilize AI within their business operations. It seems as though these treaties only outline how to manufacture a hammer, without addressing how it should be wielded.
Secondly, due to the rapid evolution of technology, these agreements tend to focus more on the technology itself rather than the individuals who may be impacted by it. For AI to benefit society, individuals must have control over its usage.
This article is directed towards business executives utilizing or integrating AI and emphasizes the necessity for any AI safety measures to prioritize people. It offers insights on AI protection, its significance, and best practices for ensuring AI safety in business leadership.
HOW DOES AI SAFETY FUNCTION?
AI safety ensures that no harm is inflicted on individuals or the environment. Risks include generating false information, compromising personal data, perpetuating biases, and displacing jobs.
The complexity of AI safety risks poses a significant challenge for business executives in terms of observation, assessment, and management due to the dynamic nature of the technology. Each new IoT application launch unveils fresh safety challenges.
Moreover, professionals often struggle to identify the root cause of harms stemming from the novelty of the technology. Is the harm a result of the individuals guiding the AI algorithms or the distortion of information? Did an employee lose their job due to inadequate skills or redundancy caused by AI? Managing these risks becomes arduous when the triggers are ambiguous and untraceable.
WHY IS BUSINESS CONCERNED ABOUT AI SAFETY?
While some businesses remain unaffected by AI, many industries, including high-tech sectors like software development and banking, as well as traditional sectors like agriculture, mining, and manufacturing, are being transformed by AI. AI is reshaping various aspects of business operations, from production and internal processes to retail and customer interactions.
Business leaders are apprehensive that failing to promptly adopt AI within their organizations will result in a loss of competitive edge. Consequently, professionals worldwide are striving to navigate this transition effectively.
However, improper implementation of AI exposes companies to safety risks, adverse reactions from employees or customers, and reputational damage. Thus, it is imperative for all businesses to integrate AI securely.
EFFECTIVE AI SAFETY STRATEGIES
When I initially delved into writing this article, my first query to ChatGPT4 was regarding the optimal methods for deploying AI correctly. While it swiftly provided advice for business leaders, the recommendations were quite conventional.
According to ChatGPT4, business executives can enhance management, assess challenges, mitigate risks, maintain transparency in their operations, and conduct audits. These guidelines are applicable across various industries, including financial reporting and automotive manufacturing, to bolster security measures.
To ensure proper AI deployment, people and the environment must be at the core of the management protocols. Artificial health hazards often sideline the well-being of individuals and the planet. Given the rapid evolution of AI technologies, it is crucial to identify emerging challenges promptly and adapt AI management processes accordingly.
Outlined below are three guiding principles:
1. Utilize AI to address challenges, not merely to explore opportunities.
Business professionals face pressure to integrate AI into their workflows, driven by the fear of falling behind competitors or the allure of potential advantages.
When I probed executives with substantial portfolios about the issues they aimed to tackle with AI, they struggled to provide a candid response. While one executive cited cost reduction, another mentioned gaining a competitive edge. However, these are aspirations, not issues. By reframing the question to focus on the problems AI could solve within their business, they began identifying pertinent challenges. It’s akin to searching for a hammer without a nail.
Deploying new technologies without a clear problem to solve can lead to unintended consequences. Companies inadvertently cause harm when their enthusiasm for opportunities blinds them to the associated costs.
AI should be viewed as a potent tool among many, rather than a standalone solution. For instance, pension fund managers must first identify existing research challenges within their department to enhance research capabilities effectively. Deploying AI without addressing underlying issues may exacerbate existing problems.
2. Integrate AI into the company’s corporate social responsibility commitment.
Many businesses have established Corporate Social Responsibility (CSR) departments tasked with conscientiously managing the company’s operations, considering factors like carbon emissions, diversity initiatives, and ethical labor practices.
However, technical or development departments often implement AI without substantial input or oversight from CSR departments. Companies seeking to deploy AI responsibly should involve CSR departments to ensure that knowledgeable individuals assess the impact of AI processes on people and the environment.
When AI deployment is solely driven by commercial motives, organizations risk making errors that could tarnish their reputation. For instance, Google faced severe criticism when its Head of AI Governance, Jen Gennai, advocated for relaxing AI safety standards in favor of using AI for research rather than product development. Involving the CSR department would have facilitated a better understanding of the implications of Gennai’s directives in alignment with the company’s overall responsibility strategy.
By treating AI with the same level of scrutiny as other strategic decisions impacting individuals and the environment, companies are more likely to address emerging security challenges thoroughly.
3. Embed AI safety into the corporate culture.
If AI safety is not a top priority for executives and employees, it may take a back seat to other business concerns. Gupta emphasizes that safety measures can be easily overlooked in pursuit of financial gains. In environments where AI companies offer lucrative salaries and operate at high burn rates, the cultural mission might be overshadowed by business objectives.
To ensure careful management of AI applications, business leaders must consistently reaffirm the company’s commitment to people and the environment. Incorporating AI safety into the organizational culture can take various forms, such as:
- facilitating discussions on AI safety in senior management and board meetings,
- establishing comprehensive AI teams comprising social experts and software engineers, and
- encouraging AI teams to participate in professional development programs and workshops focusing on cutting-edge AI safety practices.
These initiatives underscore the importance of AI security and enable businesses to swiftly adapt to the evolving AI landscape.
THE IMPERATIVE OF AI SAFETY
The world is only witnessing the initial stages of AI advancement. While the future remains uncertain, there is a consensus that more powerful, predictive models are on the horizon.
Unless business leaders address AI safety promptly, these innovations could pose significant challenges for humanity. Conversely, well-managed artificial general intelligence holds the potential to address critical global issues such as climate change, biodiversity loss, and public health.
For AI to be truly beneficial, it must be developed and utilized with a focus on people and the planet.
