Out of 90 percent product risk assessments, CEO Mark Zuckerberg announced that the tech giants plan to automatically use artificial intelligence and it will apply to all of its platforms, including Facebook, Instagram and WhatsApp. The idea behind this move is to simplify the operations and to fast track the deployment of new features.
While Meta previously carried out risk assessments involving human oversight — especially after agreeing to privacy reviews of new products with the US Federal Trade Commission in 2012 — these approaches are about to evolve. This shift towards evaluation via AI is a result of having many decisions (algorithm updates, safety tool implementation and content sharing functionality changes, etc) made by automated systems.
The company maintains that human judgement will continue to be applied to discern novel and complex cases, though there is some worry about dependence on AI for the bulk of assessments. Reduced human oversight could result in unintended consequences, critics fear such as in A.I. safety or misinformation management. Launching features faster with less scrutiny might mean we launch it with less potential for real world harm … because we don’t catch it before we launch it, said a former Meta executive.
The company argues that this alters the structure is just part of its broader plan to increase efficiency and scalability of product development. Despite saying that it has continued investing in user privacy, the company reiterates its commitment to striking a balance between innovation and responsibility.
Such an integration of AI is an emerging trend among tech companies to squeeze the much buzzed about technology in real core operational processes. With an increasing sophistication of AI systems, companies are increasingly giving the responsibility of tasks formerly reserved for human experts to perform AI tasks. Yet, the underlying necessity for adequate oversight systems to verify whether automatically generated decisions abide by ethics and the security of users is equally clear.
The bottom line is that Meta’s choice to completely automate its product risk assessments is a major step forward from its operations. Although it has the potential to increase speed and its efficiency, user privacy and safety have to be taken into account very carefully. In this transition the company has to find the right balance between innovation and accountability.
