The AI Trust Gap
The black box nature of many AI modelsoften prevents people fromtrusting intelligent machines.As we move forward, it’s crucial to prioritise the development of AI systems that are not only powerful but also trustworthy and transparent
The tech industry has poured tens of billions into developing new AI models, with leading players like OpenAI seeking trillions more in investment. The goal is to steadily demonstrate better AI performance and close the gap between human and machine capabilities. Interest in AI, building since last year, will push a 10% increase in data centre system spending this year, driving worldwide IT spending to $5.06 trillion, said John-David Lovelock, distinguished vice president analyst at Gartner.
NASSCOM reports that nearly 90% of companies plan to boost spending on top digital technology priorities in 2024, a significant increase from past years. It is predicted that over the next 6-12 months, the majority of investments will focus on AI and machine learning technologiessuch as generative AI, intelligent automation, and big data analytics.
But there is another critical gap that deserves equal, if not higher, priority – the AI trust gap. This gap refers to the persistent risks, both real and perceived, that prevent people from being willing to entrust machines with tasks that would otherwise be handled by qualified humans.
The Trust Gap Concerns
According to a recent HBR article, this trust gap spans a range of concerns, from disinformation and safety/security risks to ethical issues, bias, and the black box nature of many AI systems. And it’s a major problem – a 2023 survey found that between 37.8% and 51.4% of AI and machine learning experts placed at least a 10% probability on scenarios as dire as human extinction resulting from unsafe AI.
The trust issues go even deeper. 85% of internet users now worry about their inability to spot fake content online, a serious problem given the rise of AI-aided deepfakes in recent elections from Bangladesh to Moldova. Businesses are also grappling with a litany of AI risks, with the majority of experts seeing a high likelihood of AI systems being “jailbroken” to follow illegal commands.
The problem is that no matter how advanced AI becomes, the trust gap will remain a permanent fixture. Efforts to improve transparency, enforce ethical guidelines, and mitigate biases will only provide partial remedies. The black box nature of many AI models, the context-dependent nature of ethical dilemmas, and the inevitability of some biases mean the trust gap will persist.
This has major implications. First, AI adopters – consumers, businesses, policymakers – will always have to traverse this trust gap. Second, companies must invest in understanding and addressing the specific risks driving mistrust in their applications. A recent Cornell study, for example, found that a New York law requiring employers to audit their AI hiring tools for bias was largely toothless.
And third, pairing humans with AI will be essential, as we will always need humans to guide us through the trust gap. As one expert noted, “the industry has spent tens of billions in creating AI products, such as Microsoft Copilot. It’s time to also invest in the human alongside: the pilot.”
The lesson is clear. The industry has spent billions creating AI products, but it’s time to also invest heavily in the human component. Preparing humans to recognise the causes of the AI trust gap, accept its permanence, and learn how to effectively oversee and complement AI systems is crucial. Only then can we realise the full potential of AI while maintaining public trust.
A Path to Trust and Interconnectedness
As we stand at the threshold of 2024, the world is poised to witness a significant shift in the way we interact with artificial intelligence (AI). The advancements in large language models (LLMs) like GPT have opened doors to new possibilities, but it’s crucial that we address the trust issues that arise from these powerful tools.
The integration of LLMs with sensors and actuators will mark the beginning of a new era where AI systems interact with the physical world, controlling everything from thermostats to industrial processes.The ease with which AI can now interact with humans has led to concerns about job displacement and the potential for AI to manipulate public opinion.
However, the real challenge lies in ensuring that these systems are designed and used ethically, taking into account the potential impact on individuals and society. The trust deficit in AI is multifaceted, encompassing issues such as disinformation, safety and security, the enigma of the “black box,” ethical dilemmas, bias, instability, hallucinations in large language models, unforeseen risks, potential employment displacement and social disparities, environmental consequences, industry consolidation, and government intervention.
To bridge this trust gap, it’s essential that we empower, train, and include humans in managing AI tools. This approach will not only ensure that AI systems are designed and used ethically but also provide a framework for addressing the complex challenges that arise from their integration into our daily lives.
As we move forward, it’s crucial that we prioritise the development of AI systems that are not only powerful but also trustworthy and transparent. The path to achieving this lies in recognising the interconnectedness of all systems and the need for a collaborative approach to governance and economy. In 2024, we will take the first steps towards this future, and it’s essential that we do so with a deep understanding of the challenges and opportunities that lie ahead. The future of AI is not just about the technology; it’s about how we choose to use it to create a better world for all.