Organoid Intelligence Credit:www.embs.org
Introduction
With better technology and a more nuanced understanding of the brain’s physiology, researchers are building organoids that increasingly mimic the structure and function of the human brain. Scientists have also developed organoids that combine several types of brain cells that are capable of mimicking various regions of the brain and can mature through numerous stages of human development. Research on brain organoids is promising, but certain technical and scientific limitations still need to be resolved. For one, organoid research is not always reproducible. Sometimes cells grow and thrive, while others on the same protocol starve and die. More research will be required to resolve this mystery.
Differentiating Human Brains from Intelligent Machines
Human brains are slower than intelligent machines, for example when processing simple mathematical equations is required. But they far surpass machines in processing complex information as brains deal better with few or uncertain data. Brains can perform both sequential and parallel processing whereas computers can do only the former, and they outperform computers in decision-making on large, highly heterogeneous and incomplete datasets . The emerging field of so-called ‘organoid intelligence’ aims to leverage the extraordinary biological processing power of the brain. Both biological learning and machine learning build internal representations of the world to improve their performance in conducting tasks. However, fundamental differences between biological and machine learning in the mechanisms of implementation result in two drastically different efficiencies. First, biological learning uses far less power to solve computational problems. For example, a larval zebrafish navigates the world to successfully hunt prey and avoid predators using only 0.1 microwatts. A human adult consumes 100 watts, of which brain consumption constitutes 20%. In contrast, clusters used to master state-of-the-art machine learning models typically operate in the kilowatt range. These observations have created high expectations for biological, brain-directed computing as an alternative to silicon-based processing. However, realizing the potential of biocomputing has proved challenging, and most research remains in its infancy. Organoid intelligence describes an emerging field aiming to expand the definition of biocomputing toward brain-directed computing, for example to leverage the 3D human brain organoids to memorize and compute inputs. To learn from and harness the computing capacity of these organoids presents a major research challenge in advancing AI to a new frontier. While AI aims to make computers more brain-like, research in organoid intelligence explores how a 3D brain cell culture can be made more computer-like.
The Contribution of Neuroscience
Since the inception of AI research midway through the last century, the brain has served as the primary source of inspiration for the creation of artificial systems of intelligence. This is largely based upon the reasoning that the brain represents proof of concept of a comprehensive intelligence system capable of perception, planning and decision making and therefore offers an attractive template for the design of AI-systems. Classically, our definition of intelligence has largely been based upon the capabilities of advanced biological entities, most notably humans. Accordingly, AI-research has primarily focused on the creation of machines that can perceive, learn and reason, with the overarching objective of creating a system capable of Artificial General Intelligence (AGI) that can map human intelligence. It is not surprising that scientists, mathematicians, and philosophers engaged in AI-research have taken inspiration from the structural, and functional properties of the brain. Since at least the 1950s, attempts have been made to artificially model the information processing mechanisms of neurons. This began with the development of so-called perceptrons, a highly reductionist model of neuronal signalling, in which an individual node of an Artificial Neural Network (ANN) receives weighted inputs that produce a binary output until the summation of inputs reaches a threshold. By the late 1980s, the development of multilayer neural networks and the popularization of backpropagation had solved many of the limitations of early perceptrons. ANNs were now able to dynamically modify their layers, giving rise to a new generation of applications including image and speech recognition. A major advantage of biologically plausible AI is its usefulness for understanding and modelling information processing in the brain. Brain mechanisms can be thought of as an evolutionarily-validated template for intelligence, honed over millions of years for adaptability, speed and energy efficiency. As such, increased integration of brain-inspired mechanisms may help to further improve the capabilities and efficiency of AI and further strengthen the successful partnership between AI and neuroscience in the years to come.
Learning Advantages of Organoids
Organoids are three-dimensional tissue cultures commonly derived from human pluripotent stem cells. What looks like a clump of cells can be engineered to function like a human organ, mirroring its key structural and biological characteristics. Under the right laboratory conditions, genetic instructions from donated stem cells allow organoids to self-organize and grow into any type of organ tissue, including the human brain. Although this may sound like science-fiction, brain organoids have been used to model and study neurodegenerative diseases for nearly a decade. Emerging studies now reveal that these lab grown brain cells may be capable of learning. Researchers speculate that this so-called ‘intelligence in a dish’ may be able to outcompete artificial intelligence which is broadly defined as the ability to acquire, store and apply information. Executing any task requires some level of intelligence irrespective of consciousness or self-awareness. An AI chatbot like ChatGPT, for example, can respond to its users in real time with curated responses, but the extent of its intelligence is bound by data-based algorithms. Computers inherently cannot ‘think’, or ‘feel’ on their own. Similarly, brain organoids can learn to perform tasks, but there is no evidence that they are capable of consciousness. Intelligence alone is not sufficient for the subjective feeling of consciousness. While AI-machines can process information much faster than the human brain, AI technology is currently limited to sequential processing, and therefore, only excels at tasks that can be done in chronological order, such as mathematical computations. In contrast the human brain has the unique ability to perform parallel processing, allowing us to analyse multiple pieces of information simultaneously. Even with something as simple as the sense of vision, our brains can identify the colour, shape, position, and reflective speed of an object instantaneously. When encountering unfamiliar or changing information, human intelligence fairs far better than intelligent machines. Tests conducted by researchers confirm that individuals already at the age of four need far fewer trials to learn a new task compared to conventional AI-systems.
Organoid Intelligence: Future Directions
As this emerging field continues to advance, several ethical questions have yet to be answered. Currently, there is no evidence that these cells have a consciousness, or the ability to ‘think’ or ‘feel’ on their own, but as these models are scaled up, we cannot be sure that they never will. Consciousness is far more complex, and we may ever be able to understand its mystery. It is possible, however, that research will eventually encounter the so-called Greely Dilemma: By developing more realistic and ethical methods for studying the brain, we may create new models that are eligible for moral and ethical protections as well. As these models become more ‘human-like’, researchers will be forced to define what makes a person, a person. Even if these 3D brain cultures never become sentient, there is still the question of who owns the intellectual brain property that they produce. This is a question that has already been asked in the application of AI. The fact that the stem cells used to generate brain organoids come from donor volunteers further complicates this ethical question. Do those volunteers retain the rights for anything created using their genetic code? While the field of organoid intelligence seems promising, ongoing research will require strong boundaries to enforce the ethical and socio-economic issues.
Conclusion
In theory, artificial brains are man-made objects that are just as intelligent, creative and as self-aware as humans. However, no such machine has yet been built and to some scientist’s view artificial brains will never reach the functionality level of biological human brains. These critics argue that there are aspects of human consciousness or expertise that cannot be simulated by machines. To others it is only a matter of time until the functionality of artificial brains will be equal to that of human brains. Based on the exponential growth of scientific knowledge this could happen within decades, raising the question as to how humanity will adapt to this evolutionary milestone.