By Professor Louis C H Fourie
Artificial Intelligence (AI) currently drives many real-world applications, ranging from facial recognition to language translators and personal assistants.
Simultaneously, companies across industries are increasingly harnessing the power of AI in their operations to improve productivity, growth and innovation. As the technology progressed, AI gradually started to take over more and more mundane tasks, thus eliminating human effort and the constant risk of error inherent to humans.
AI does the heavy thinking
The popularity of the Internet of Things and the myriad sensors and autonomous devices are constantly producing massive amounts of data. Interpreting this huge amount of data and turning it into actionable insights is just not possible for human beings. People thus started to employ AI and the superior computational and problem-solving capabilities of machines to assist with the examination of the enormous amounts to make better- informed decisions. Humans are thus using AI to do the heavy thinking for them, for example:
Automate the complicated analysing of complex data to find trends, patterns, and associations. Monitor and study real-time data, autonomously adjusting behaviour with little need for supervision. Discover operational inefficiencies. Increase accuracy and efficiency. Predict future outcomes based on observed historical trends.Inform fact-based decisions.
Execute plans. Constantly learn and improve through machine learning without being explicitly programmed by humans.
AI and unsolved challenges
By analysing and understanding the data better, we are able to generate improved responses to confront day-to-day issues. However, the steady progress in basic AI tasks in recent years and the steady stream of advances in AI and in particular machine learning (a technological subset of AI that enables computers to autonomously adjust when exposed to new data), have opened up a multitude of promising opportunities for AI applications.
It seems that we may be on the verge of creating AI that will be capable of finding solutions to the world’s most pressing challenges.
Eric Schmidt, former executive chairperson of Alphabet (previously Google) and Demis Hassabis, the chief executive of Deepmind (a division within Google doing ground-breaking work in machine learning), for quite some time have been saying that powerful computers and developments in AI will in future solve major global challenges that humans have been unable to solve until now..
Some of these complex problems that may in future be solved by AI are the early detection of pandemics, rapid case diagnosis, climate change, poverty, food security, energy efficiency, disaster prediction, fraud detection, and early conflict detection.
Just imagine having AI sifting through thousands of documents and simulations electronically to find cures for many of life’s serious challenges – continuously learning as it continues.
Solving a complex biological problem
Proteins are large complex molecules, made up of chains of amino acids. Scientists have been struggling for almost half a century with the problem of “protein folding” – the mapping of the three-dimensional shapes of the proteins that are responsible for diseases from cancer to Covid-19.
Proteins are the microscopic mechanisms that drive the behaviour of viruses, bacteria, the human body and all living things. What a protein does largely depends on the unique 3D structure. Understanding which shapes proteins are folding into is widely known as the “protein folding problem” that has stood as a major challenge in biology for half a century.
The ability to predict this structure would provide a greater insight into what the protein does and how it operates. The development of treatments for diseases or finding enzymes that break down industrial waste depends on the folding of the proteins and its amino acid sequence.
A few days ago, the company DeepMind announced on its blog that its AI system, AlphaFold, solved this half-a-century old challenge of “protein folding” by computationally predicting protein structures with incredible speed and precision. This solution is a major advance in biology.
What makes this breakthrough so remarkable is that it happened many years earlier than expected. In 1969 Cyrus Levinthal stated that it would take longer than the age of the known universe to enumerate all possible configurations of a typical protein (10 300 conformations) by brute force calculation. Yet in nature, proteins fold spontaneously, often within milliseconds – a dichotomy often called the Levinthal’s paradox.
Through the use of new deep learning architectures, DeepMind achieved unparalleled levels of accuracy of 92.4 percent, which rivals the accuracy level of physical experiments.
They regarded a folded protein as a “spatial graph” and created an attention-based neural network system to interpret the structure of the graph.
Real-world impact
This incredible biological breakthrough clearly illustrates the impact that AI can have on scientific discovery, its potential to solve complex challenges and its potential to radically accelerate progress in the understanding of many fundamental fields of the world. Until now, about 200 million proteins are known, but only a small proportion has been unfolded to understand how they operate. Current unfolding techniques require expensive equipment and often take years or decades of experimentation to complete.
In particular the protein structure predictions could contribute to the understanding of specific diseases by, among others, identifying proteins that have malfunctioned. These insights could assist the development of improved medicine for the treatment of specific diseases, unlock the mysteries of the human body, as well as speed up the development process of new drugs and the repurposing of existing medications as a cocktail to treat new viruses and diseases.
AI can thus dramatically change the fight against diseases if they can determine how the drugs will bind or physically attach to the protein molecules to alter its behaviour.
For example, Andrei Lupas from the Max Plank Institute for Developmental Biology in Germany has been struggling to determine the shape of a particular protein in a tiny bacteria-like organism called an archaeon.
Since the protein straddles the membrane of individual cells, even after a decade he could not determine the shape. With the help of AlphaFold he solved the problem in half an hour.
Although the breakthrough might be too late to make a significant impact on the coronavirus, fast and accurate protein structure prediction could be very useful to accelerate future pandemic response efforts.
Earlier this year, DeepMind was able to predict several protein structures of the Sars-CoV-2 virus, including a protein of which the structures were previously unknown.
AlphaFold may also in future assist in the development of new vaccines for viruses.
Some researchers also believe that the DeepMind AI system could help scientists gain a better understanding of genetic diseases such as Alzheimer’s or cystic fibrosis.
The future
The AlphaFold AI system by DeepMind has clearly demonstrated the immense potential for AI as a tool in fundamental discovery. AI technology, in particular neural networks (a mathematical system modelled on the network of neurons in the human brain), enabled machines to perform numerous tasks that were once beyond their reach, as well as the reach of humans.
However, just as Nobel Prize in Chemistry winner Christian Anfinsen’s hypothesis that a protein’s amino acid sequence determines its structure laid out a challenge to computationally predict the 3D structure of the protein that was far beyond the reach of science at the time, there are still many challenges that remain to be solved.
But the progress made by DeepMind brings hope that AI could become one of humanity’s most useful tools in expanding the frontiers of science and solving the unsurpassable problems of the world.
Eventually AI will be found in every industry on planet Earth. Although much development and more breakthroughs are still needed, AI can indeed help to solve a number of global issues that humans were unable to solve until now.
The American author, Helen Keller, once said: “Optimism is the faith that leads to achievement. Nothing can be done without hope and confidence.” Who knows, one day AI might just find the cure to cancer and solve the poverty and hunger issues of the world.
Professor Louis C H Fourie is a futurist and technology strategist
BUSINESS REPORT