Imagine a world without computers. Now imagine a world without artificial intelligence (AI). I expect it is easier to imagine the second of these scenarios. But, AI has crept into our everyday lives. From unlocking your phone to personalised ads, a world without AI is becoming increasingly more difficult to imagine.
The rapid development of computing technology is evident, with the launch of AI systems like ChatGPT by OpenAI into the public domain.
But, there are barriers to the technological revolution. Barriers which researchers at John Hopkins University believe can be overcome using biology. Earlier this year, the research group presented a paper laying out their plans for biological computers running on human brain cells.
The ceiling to the technology boom
Though it may seem like the technological revolution won’t be slowing down anytime soon, computers are limited when it comes to energy and data.
- Advertisement -
A key member of the research team is Thomas Hartung, professor of environmental sciences at John Hopkins Bloomberg School of Public Health and Whiting School of Engineering. Professor Hartung says: “Computing and artificial intelligence have been driving the technology revolution, but they are reaching a ceiling”.
Computers can process simple information at speeds unfathomable to the human brain. Just ask ChatGPT to write you a song, or explain rocket science and you will be amazed by how quickly you will receive an answer.
However, even the best AI doesn’t work quite like the brain.
There are two key differences when comparing learning by human brains and human-made computers, known as machine learning. Firstly, computers use a lot more energy to carry out a task. The input of power into the most advanced supercomputers is huge relative to the power a human needs to produce the same output.
Secondly, computers need much more data to learn how to complete a task. Whereas human brains outperform AIs in decision-making and can work with information from small or incomplete data sets.
Practically, this means you can show a human who has never seen cats or dogs before a few images, and they will quickly learn how to tell the difference. An AI, however, will need to observe thousands of images of cats and dogs to learn to tell them apart.
So, the human brain currently surpasses computers by using less energy and data to learn and complete computations. In other words, brains are more powerful and data-efficient. As computers become more advanced, their relative inefficiency in energy and data is thought to hinder their development.
However, the research group believe that using “biological hardware”, meaning brain cells, to make computers will overcome these barriers (1).
- Advertisement -
What is Organoid Intelligence?
The term “biocomputer” conjures up images of a brain bobbing in a jar, plugged into various electrodes and devices. But, images from science fiction are far from what researchers have in mind. The paper by researchers at John Hopkins University outlines the plan to use brain organoids, combined with innovations in biotechnology and bioengineering, to build computers based on biological materials.
While this may seem like the stuff of science fiction, organoids are already being used in research.
Organoids are tiny, lab-made tissues representing a larger organ by mimicking structure and function on a cellular level. These tiny tissues (the size of a pen dot, 1) made from stem cells help scientists study diseases without using organs from animals or humans. Brain organoids contain around 50,000 neurons. This may seem like a lot, but it is actually a mere 0.00005% of the brain’s 100 billion.
So, brain organoids are tiny representations of the brain, retaining the basic functions of learning and memory. They are 3D biological structures with a high density of neurons, high levels of supportive cells, and the expression of key learning genes. This mixture of biological matter is what researchers term ‘intelligence-in-a-dish’, perhaps not so far from that floating brain in a jar.
Just as AI describes the intelligence of (silicon-based) computer systems, the group coined the term organoid intelligence (OI) to describe the computing intelligence of biological brain organoids. And the development of this intelligence when interfaced with technology.
An organoid itself isn’t a bio-computer. Think of organoids as the building blocks used to build the bio-computer, just as the brain has different regions that fit together to perform the overall function. While we have basic brain organoids, the next stage involves developing a biocomputer optimal for learning. The researchers have a vision of a complex network of brain organoids, computers/AI, data warehouses and ‘sensors’ to feed information into the biological system (3).
The organoids will then be ‘trained’ to learn different tasks, taking advantage of the brain’s innate learning capacity, and connected to computers to perform a computational task.
Hartung says, “We anticipate OI-based biocomputing systems to allow faster decision-making, continuous learning during tasks, and greater energy and data efficiency.” (3)
- Advertisement -
An ethical approach
Beyond all the technical challenges of OI, importantly, the team recognise the imperative of ethics. The aim is not to replicate human consciousness but to utilise the amazing learning and cognitive capacities of the brain. Nonetheless, we face ethical challenges when modelling the brain. This was laid out in Greely’s dilemma.
The dilemma is as follows: Models of the brain are used in research to avoid ethical issues of using human brains. But, if the models become too good they will become deserving of the ethical treatment that motivated researchers to use models in the first place.
Considerations like Greely’s dilemma created the imperative of an ‘embedded ethics‘ approach to OI. In other words, a team of researchers, ethicists, and the public will discuss issues as the research develops, but in a framework of anticipated ethical issues laid out from the onset of the paper.
With this final strand, the research group hope to establish a plan that can utilise the capacity of the human brain to overcome barriers to the technological revolution, while retaining ethical responsibility. However, as the field of OI is born, some key questions remain:
- Are there other methods of increasing computational energy and data efficiency without ethical risk?
- Will OI replace AI?
- And how could OI change how we think about human learning and consciousness?
Hartung admits that this dream is something of the future and that it could take decades before we develop a biocomputer as intelligent as a mouse. But researchers agree on one thing – the future starts now.
