Several Employees from Companies Specializing in Artificial Intelligence Express Desire for ‘Right to Notify’ Regarding Risks
On June 4, a group of both current and former staff members from OpenAI, Google DeepMind, and Anthropic released a statement requesting safeguards for whistleblowers, increased transparency concerning risks, and the promotion of a culture that embraces constructive criticism within the leading artificial intelligence companies.
The missive, known as the Right to Warn letter, sheds light on the inner mechanisms of the select high-profile organizations situated in the spotlight of artificial intelligence innovation. Of note is OpenAI’s distinctive position as a not-for-profit entity striving to “navigate substantial risks” related to theoretical “general” AI.
For enterprises, this initiative arrives amidst a growing advocacy for the adoption of automated AI tools, emphasizing the critical importance for technology decision-makers to establish robust guidelines governing the usage of AI.
The Right to Warn letter urges pioneering AI enterprises to abstain from retaliating against whistleblowers and more
The key demands outlined in the letter are:
- Prohibiting advanced AI companies from implementing agreements that restrict criticism directed towards these companies.
- Establishing a confidential channel through which employees can voice concerns about risks to the companies, regulatory bodies, or independent entities.
- Promoting “a culture that embraces open criticism” regarding risks, while accommodating the protection of proprietary information.
- Putting an end to reprisals against whistleblowers.
This development follows closely after disclosures of stringent non-disclosure agreements for departing staff at OpenAI, around two weeks ago. Allegedly, breaching the non-disclosure and anti-defamation agreement could result in forfeiting the rights of employees to their vested equity in the organization, potentially surpassing their salaries. On May 18, OpenAI’s CEO Sam Altman expressed being “embarrassed” by the implications of revoking vested equity from employees and pledged to revise the agreement.
All current employees from OpenAI who endorsed the Right to Warn statement did so anonymously.
What kinds of risks associated with automated AI are highlighted in the letter?
The letter draws attention to potential hazards arising from automated AI, identifying risks that span “from exacerbating existing disparities, to manipulation and disinformation, to relinquishing control over autonomous AI systems that could lead to human extinction.”
OpenAI’s stated mission has always revolved around crafting and safeguarding artificial general intelligence, sometimes referred to as AGI. This concept refers to theoretical AI surpassing human intelligence or capabilities, evoking thoughts of sentient machines and humans playing subordinate roles, as seen in science fiction. Some critics of AI argue that these apprehensions divert attention from more immediate challenges at the crossroads of technology and society, such as the misappropriation of creative content. The authors of the letter encapsulate both existential and societal perils.
How could internal apprehension from the tech sector influence the accessibility of AI solutions for enterprises?
Businesses not positioned as frontier AI entities but evaluating the incorporation of automated AI might leverage this statement to review their AI deployment protocols, secure and trustworthy evaluation of AI products, and the traceability in data sourcing throughout the use of automated AI.
SEE: Companies should thoughtfully forge an AI ethics policy tailored to their business objectives.
Juliette Powell, co-author of “The AI Dilemma” and an ethics professor at New York University specializing in artificial intelligence and machine learning, has extensively studied the outcomes of employee protests against corporate policies over the years.
“Cautionary open letters from employees alone may not yield significant impact without public support, which, when coupled with media influence, garners enhanced leverage,” she conveyed via email to TechRepublic. For instance, Powell suggested that authoring editorials, exerting public pressure on the boards of companies, or withholding investments in frontier AI ventures would potentially be more efficacious than merely endorsing an open letter.
Powell referenced a previous plea for suspending giant AI endeavors from last year as another instance akin to this initiative.
“I believe the likelihood of major tech firms consenting to the conditions of such letters – AND UPHOLDING THEM – is as probable as holding computer and systems engineers accountable for their creations, akin to how civil, mechanical, or electrical engineers are held responsible,” Powell remarked. “Therefore, I perceive that a communication like this would not significantly impede the accessibility or utilization of AI solutions for enterprises.”
OpenAI has consistently included acknowledgment of risks in its persuit of increasingly sophisticated automated AI, suggesting that numerous companies may have deliberated the merits and demerits of adopting automated AI products for their operations. Inter-organizational discussions about AI application guidelines could integrate a policy endorsing a “culture of open criticism.” Executives could contemplate instituting safeguards for employees discussing potential risks or investing solely in AI solutions deemed to promote responsible socio-ethical and data governance ecosystems.
About Author
