Image: Supply
“Computers deserve to be called intelligent if they can deceive humans to believe they are human.” Alan Turing
We have come a long way since the 1950s, especially the beginning of modern AI in the last few years. I think we are at a turning point now that AI changes the way research is done and the way the industry interacts with these technologies. Politics and society need to coordinate and confirm that AI is used in an ethical and safe manner and address privacy concerns. There are many possibilities with AI, but there are still many issues and concerns. If you can deal with these things, you can pioneer the good things about AI.
Alan Turing (1912-1954) is a British mathematician and computer scientist, and is widely known as the father of theoretical computer science and AI. He made many notable contributions. For example, he introduced the concept of a theoretical computing machine, also known as the Turing machine. He worked on early computer design using the National Physics Laboratory and later worked at the University of Manchester, where I was based. He takes on a pioneering job, which continues to influence modern computer science. He also developed the Turing test, which measures the machine’s ability to demonstrate intelligent behavior that is indistinguishable from that of a human.
Turing Test: Why is it relevant?
The Turing test is still in use today. Turing introduced it as a test of what is known as a mimicry game in which human interrogators interact with two hidden entities (one human and another machine) through text-based communication similar to ChatGpt. The interrogator cannot see or hear participants, and must rely solely on the conversations in the text to determine whether it is a machine or a human. The purpose of the machine is to generate responses that are indistinguishable from human responses. Human participants aim to convince the interrogator of her/his humanity. If the interrogator cannot reliably distinguish between a machine and a human, the machine is said to have passed the Turing test.
It sounds very simple, but it’s an important test as it’s a classic benchmark for assessing AI. However, there are also criticisms and limitations on the test. Marking Alan Turing in 2024, we can say that AI is nearing passing the Turing test, but we are not there yet.
In a recent paper, ChatGpt said he passed the Turing test. ChatGpt is a natural language processing model that generates answers to questions that appear to be human responses. Some say that ChatGpt has passed the Turing test and certainly for the short conversations, ChatGpt does a very good job. However, as I have a longer conversation with ChatGpt, I notice that there are some flaws and drawbacks. So, ChatGpt is probably the closest we can pass the Turing test at this time.
Many researchers and companies are working on improving the current version of ChatGPT. I want to make sure I understand what the machine generates. At the moment, ChatGpt produces a set of words that are suitable for dealing with a particular query, but they do not understand the meaning of these words. If ChatGpt understands the true meaning of a sentence, and if it was done by contextualizing a particular response or query, then I think we are in the position that we have passed the Turing test. I’ve always wanted to go through this stage, but I hope we’ll reach this point in a few years, perhaps around 2030.
The University of Manchester addresses many aspects of AI in healthcare. It starts with drug discovery. Can you find a drug that is more potent than drugs, has fewer side effects, and is ideally cheaper to manufacture than currently available drugs? Use AI to guide you through the search space for different drug combinations. And, for example, AI will tell you which drugs to combine, and which dosages to take.
It is also working with the UK National Health Services to come up with a fair reimbursement scheme for hospitals. In some cases, we use what is called sequential decision making. The other uses a technique based on a decision tree. So, we will use different methods to consider different applications of AI within healthcare.
The specific area of cybersecurity I’m working on is secure source code. This is the way we tell our computers what to do and is one of the basic levels of human interaction with our computers. If the source code (a set of instructions) is of low quality, it can open up security vulnerabilities that hackers could exploit. Using verification techniques combined with AI, it scans the source code to identify and fix various types of security issues. By doing that, we have shown that it will improve the quality of the code and improve the resilience of the software. We generate a lot of code and want to make sure the code is secure, especially for businesses in high interest sectors such as healthcare, defense, and finance.
Sports ai
Creativity and sports have a lot of scope and potential for AI. In soccer, there is data on match action. It is positioning for people with the ball, people with the ball, and players. It’s really big data, you can look at past performances and player styles, analyze and use the data to adjust your strategy when playing a particular opponent. Due to the amount and complexity of data, this is extremely difficult without AI.
They also consider music education and create virtual music teachers to help people learn better about musical instruments. You can use AI in conjunction with other technologies such as Virtual Reality and Augmented Reality to project tutors. Wearing VR goggles will allow you to actually interact with your tutor. This is extremely innovative and has the potential to open music to everyone on the planet.
At this point, AI is very good at performing certain tasks and is at a stage where it is making very good progress with AI Generals. AI is capable of acting and interacting in the same way as humans. This is a game changer made possible by ChatGpt and other examples. This technology is being used in the industry for all-new business ideas that we don’t think about.
The vision and strategy of AI is extremely important. The UAE National Strategy for AI 2031 is a very good example of an ambitious vision covering education, reskills and research investments.
This strategy also focuses on the development of ethical AI to ensure that AI is used ethically and safely and reduces privacy concerns. This strategy has all the components needed to be successful and I think we can all learn a lot from this approach.
The writer is Professor of Applied Artificial Intelligence and Associate Dean of Business Engagement, Civic and Cultural Partnerships (Humanities) at the Alliance Manchester Business School at the University of Manchester.
Read the AI Building Trust: A Journey to the Future of Digital Cognition in the United Arab Emirates
