Image: Provided
“A computer would deserve to be regarded as intelligent if it could trick a human into thinking it was human.” – Alan Turing
Since the emergence of modern artificial intelligence in the 1950s, there has been significant progress, particularly in recent years. We are now at a crucial juncture where AI is transforming research methodologies and altering how industries engage with these technologies. Consequently, politics and society must adapt to ensure AI is ethically utilized and privacy issues are tackled. Although AI holds tremendous promise, various challenges and concerns still exist. If we can address these, a bright future with AI awaits.
Alan Turing (1912 – 1954), a renowned British mathematician and computer scientist, is often recognized as the father of theoretical computer science and AI. He made numerous important contributions, including the introduction of the Turing machine concept, which laid groundwork for contemporary computer science. Turing was also involved in the design of early computers at the National Physical Laboratory and at the University of Manchester, where I am currently based. His pioneering efforts continue to impact modern computing. Additionally, he developed the Turing test, which assesses a machine’s ability to demonstrate intelligent behavior indistinguishable from a human’s.
The Importance of the Turing Test
The Turing test remains relevant today. Turing proposed it as the imitation game, where a human evaluator interacts with two hidden participants—one human and the other a machine—using text-based communication, reminiscent of ChatGPT. The evaluator has no visual or auditory access to the participants and must base their judgment solely on the text exchange to determine whether they are conversing with a machine or a human. The machine’s goal is to craft responses that are indistinguishable from those of a human. Conversely, the human participant aims to persuade the evaluator of their humanity. If the evaluator cannot consistently tell apart the machine from the human, the machine is said to have passed the Turing test.
While the test appears simple, it serves as an essential benchmark for evaluating AI. However, it does face criticisms and limitations. As we commemorate Alan Turing Day in 2024, I can assert that AI is edging closer to successfully passing the Turing test—but we haven’t reached that milestone yet.
A recent paper claimed that ChatGPT had passed the Turing test. As a natural language processing model, ChatGPT produces responses to inquiries that resemble human replies. Some argue that ChatGPT has indeed passed the test, particularly in brief exchanges. However, as conversations extend, certain flaws and vulnerabilities become apparent. Thus, in its current state, ChatGPT is likely our closest attempt to passing the Turing test.
Many researchers and organizations are dedicated to enhancing the current version of ChatGPT. I hope to see progress where machines can comprehend their own outputs. Presently, ChatGPT generates sequences of words that seem appropriate for specific queries but lacks true understanding of their meanings. If ChatGPT could grasp the real significance of a sentence through contextualization, we could confidently say it has passed the Turing test. I had hoped for this advancement to occur sooner, but I anticipate reaching this level by around 2030.
At the University of Manchester, we are exploring various dimensions of AI in healthcare—seeking better, more affordable, and faster treatments for societal benefit. Our work begins with drug discovery. The aim is to identify drugs with increased potency, fewer side effects, and lower production costs compared to existing options. We employ AI to navigate through a range of drug combinations, guiding us on which drugs to mix and in what dosages.
We also collaborate with the UK National Health Service to develop fairer reimbursement models for hospitals. One approach utilizes sequential decision-making, while another applies methods grounded in decision trees. Thus, we implement various strategies and examine diverse AI applications within the healthcare sector.
My area of focus in cybersecurity involves secure source code—one of the basic levels at which humans interact with computers. Poor-quality source code can lead to security vulnerabilities that hackers may exploit. We utilize verification techniques combined with AI to analyze source code, pinpointing different types of security issues and rectifying them. Our efforts have demonstrated that this process enhances code quality and bolsters the resilience of software. Given that we generate substantial amounts of code, particularly for high-stakes sectors like healthcare, defense, or finance, ensuring code safety is paramount.
AI in Sports
The potential for AI in creativity and sports is vast. In football, we analyze data regarding match actions—such as the ball’s location, possession, and player positioning. This massive dataset allows us to refine strategies against specific opponents based on historical performance and playing styles, a process that would be extraordinarily challenging without AI due to the data’s complexity and volume.
We are also investigating music education, helping individuals learn instruments more effectively through virtual instructors. By integrating AI with other technologies like virtual and augmented reality, we can create an interactive tutoring experience. With VR headsets, students can directly engage with their virtual tutors, revolutionizing music education and potentially making it accessible to a global audience.
Currently, AI excels at performing specific tasks, and we are advancing towards general AI—machines exhibiting behaviors akin to humans that we can interact with. This transformative development has been enabled by ChatGPT and similar innovations, and industries are leveraging this technology to create entirely new business ventures.
A strategic vision for AI is essential. The UAE National Strategy for AI 2031 exemplifies an ambitious framework covering education, reskilling, research investment, and the application of research findings. This strategy also emphasizes ethical AI development, ensuring its secure use and addressing privacy concerns. I believe this comprehensive approach can serve as a model for success, offering valuable lessons for all of us.
The author is a professor of Applied Artificial Intelligence and Associate Dean for Business Engagement, Civic & Cultural Partnerships (Humanities) at Alliance Manchester Business School, The University of Manchester.