From brains in a dish to intelligent machines

Are organoids the new frontier of AI? Q&A with Professor Thomas Hartung, John Hopkins

art pop illustration of organoid intelligence
Midjourney AI, Digital

ORGANOIDS are three-dimensional clusters of cells that mimic the structure and functions of an organ in the human body. These miniaturised organs provide researchers with accurate representations of human organs that can be used for disease modelling, drug discovery, and studying developmental biology. Starting with stem cells, or reprogrammed adult cells, organoids can be grown from various tissues: the liver, pancreas, intestine, kidney, lung, heart – and the brain.

Brain organoids provide a platform for testing hypotheses about brain function and development that would be otherwise difficult – practically and ethically – with living specimens. Using the ‘brain in a dish’, scientists can gain new insights into the complex interactions between neurons and the signalling pathways that govern their behaviour.

The most speculative, and headline-catching, field of organoid research is in artificial intelligence. The current generation of AI (if they can be called that) use a variety of learning networks – black boxes of complex computational logic – trained on deluges of data sets.

While ChatGPT and Midjounery have ignited popular discourse, and may sometimes seem to display brilliant flashes of intelligence, any time spent interacting with them reveals their limitations. Both are still constrained by what has been spoonfed to them, and tend towards predicatability. As things stand, there is still some debate as to whether these tools herald something revolutionary, or represent a mere upgrade in human productivity.

Biological, ‘living’ models of the human brain may enable AI development that more closely mimics the human brain’s information processing capabilities. The hope is that organoid intelligence could enable the development of sophisticated machines with more advanced cognition than anything based on computer science alone.

Below is a brief Q&A with Professor Thomas Hartung MD, a researcher at the John Hopkins University’s Department of Environmental Health and Engineering, who indulged our questions about the potential for organiod intelligence – both practical and fantastical…

Please tell us, in a paragraph, about yourself and how long you’ve been working on developing OI.

I have five professorships at Johns Hopkins, Georgetown University and University of Konstanz, Germany. They span from toxicology, pharmacology to microbiology, immunology and engineering. I am directing centre for alternatives to animal testing in the US and Europe. I have been working on rat brain organoids since 2002. After moving to Hopkins in 2009, we started humanizing this model in 2011.

In 2016, we were the first to report the mass-production of standardized brain organoids. At the time – a bit sloppily – I said they are ‘thinking’, because they are spontaneously electrophysiologically active and form neural circuits. When some people asked whether this means consciousness, I said: ‘they have nothing to think about without input and about.’

This prompted me four years ago to think, what happens when we give them input and output. Since then we started OI, which has attracted so far more than 80 researchers.

At its simplest, is organic intelligence simply growing brain cells and training them to function as a computer?

Pretty much so. Bioengineering allows to culture brain organoids with a complexity promising to realize basic cognitive functions. Combining them through multi-electrode arrays allows bidirectional communication. We are only at the beginning to find out how far this can lead to learning and memory and how this changes the organoid.

Could a bio-computer develop emotional intelligence? Consciousness?

We are certainly far away from this, but I would not exclude this. Therefore, we have a strong team of ethicists involved. At this moment, the brain organoids are tiny, about as many neuronal cells as a fly. The development of any cognition depends certainly on what input we give to the system. No emotions without feeding in emotional content. No pain without pain receptors. At this moment, the input is very simple like computer game environments.

There is a school of thought that AI, at its most advanced, could endanger the human species.  Could this ultra-intelligent organic matter become uncontrollable, science fiction-style?

I would speculate that we are decades away before such a system could even match a small mammal in its cognitive abilities. We have certainly to debate whether we would ever give such systems autonomy.

Can you elaborate on this quote below re. the applications of your research?

‘The many possible applications of this work include a new generation of biological and hybrid (biological-electronic) computing technologies, together with advances in our understanding of the physiology of cognition, learning, and memory, and the pathophysiological effects of developmental and degenerative diseases, intoxication, and infection – which in turn could stimulate drug development and other interventions.’

OI also has the potential to unlock new neuromimetic AI algorithms (with the potential to overcome current AI limitations) and aid the development of new brain-computer-interface technology.

At this moment, I see much more the opportunity to study cognition and find drugs or toxicants influencing them than for practical biocomputing. However, by learning about the human machinery of cognition, we might be able to find new computer designs.

If we can realize some of the advantages of a brain over a computer (efficiency, intuition, progressive learning, creativity, emotional intelligence etc.) there might be room for a biological component in our IT infrastructure, but this is currently science fiction.

Almost every paradigm shift in science presents us with ethical dilemmas – is this a consideration in your research?

We are very aware of the need of ethical discussions and have involved ethicists from the very beginning. This is called embedded ethics. It is fascinating to ask where sentience or consciousness for example start, or when a model might be suffering. There are also questions on how an organoid relates to the cell donor or what an informed consent needs to include to allow such research?

Could OI be applied to replicating a specific individual’s brain, along with memories and consciousness or is this firmly in the realm of science fiction?

No. Our memories are stored as connections between the neurons. I cannot imagine that we can read this out and recreate. But perhaps my imagination is too limited…