Request for opportunities

Did you know it can take up to 1,000 years for a single centimetre of topsoil to form? Soils store more carbon than the atmosphere and all of the world’s plants and forests combined. They store 1.5 Olympic swimming pools of water per hectare and enable 95% of global food production. Yet we have limited understanding of how this complex ecosystem functions – for example, only about 1% of the microorganisms found in soil have been identified so far, even though they play a major role in carbon sequestration and food security. 

In the words of Roosevelt, “A nation that destroys its soils destroys itself.” As of 2020, as much as 75% of soils had been degraded to varying degrees due to human activity. Why aren’t we doing more to engineer soil – not just to restore its health but also enhance it? What would the ideal soil system be able to do?

– Gillian Koehl, Technical Specialist for Precision Neurotechnologies

Imagine what we could do with an army of tiny robots that coordinate themselves to perform complex tasks. The world is filled with examples of tiny robots that behave just like this: insects. What’s stopping us from engineering that capability for the benefit of humanity?

Consider a hypothetical tiny robot, 2mm in each dimension. About 100 million of these could fit into a large suitcase. Each would have its own chemical power source, responding to higher-level direction (e.g. from your smartphone) that allows coordination across the colony to accomplish complex tasks. Individually, each robot would be small and expendable; together they could achieve remarkable results (consider how ants build nests and move tons of plant matter many miles).

The electronics to give the robots some autonomy and two-way communication exists. The real bottlenecks appear to be the power source and propulsion systems, as conventional motors and batteries will simply not work at the length/volume scales concerned. With world-leading experts in polymers, electrochemistry and materials engineering, the UK is well-placed to develop synthetic nano-actuation and propulsion technologies. Is it time to build molecular machines?

– Mark Symes, Programme Director for Exploring Climate Cooling

Imagine a world where living plants are online, so that signals detected by easy-to-adapt sensors can trigger biochemical responses within the organism. Data streams such as weather forecasts, physiological state of neighbouring fields, or early signs of pathogen outbreaks in other plants could be encoded in standardised genetic circuits or electrochemical signals in real time. In this world, farmers could trigger acclimations to abiotic stresses based on signals that are impossible to measure for any one plant, or forecasts that are available well before any cellular response is possible.

Synbio has enabled cellular biosensors to track concentration of molecules in real time. But these are complex to develop and only track single chemicals rather than tissue or organism-wide physiological adaptations. Plant phenotyping platforms, on the other hand, allow us to monitor adaptations at the organism level, but are not connected in any way to the plant. Combining the two allows us to imagine ‘cyborg plants’, a frameshift in the way we think about optimisation of agricultural systems. How do we take this forward?

– Fabrizio Ticchiarelli-Marjot, Technical Specialist for Synthetic Plants

Batteries allow us to swap out fossil hydrocarbons for a different set of high density fuels: metals. Metals are cheap, dense, transportable, and most importantly, they don’t vaporise into greenhouse gases upon combustion. But from a first principles perspective, we’re still doing a lousy job of leveraging metallic fuels. Lithium as a material has roughly the same specific energy as jet fuel. But even the most advanced lithium battery still comes nowhere near the practical utility of a traditional jet engine. This begs the question: how can we better leverage metallic fuels? We hear a lot about research toward better batteries and fuel cells, but we rarely hear about direct combustion. Why not? We already use aluminium powder as the fuel to boost rockets into orbit; is it so out-there to imagine we might further tame these metallic beasts without having to cage them in heavy intercalation electrodes?

Granted, burning metals may be too far fetched for aviation, but how about low-cost industrial heat – another key barrier to decarbonisation. Burning cheap, high density metals like iron produces stable oxides that in theory can be sustainably regenerated/reduced back to the original fuel. We hear a lot about the hydrogen economy; are we sure there isn’t room for an iron economy in our vision for powering the future?

– Ilan Gur, CEO

The existence of hormones has been known for more than a century, but our understanding of their effects on our bodies is in its infancy. There may be up to 200 hormones affecting our physical and mental health, but we typically only measure a handful in blood tests. Hormone imbalance is responsible for major disease burdens through endocrine disorders, particularly for those assigned female at birth, where disorders such as polycystic ovary syndrome (PCOS) affect up to 1 in 10.

Moreover, hormones are known to impact our systemic immune modulation and metabolism, linking them to a range of deadly diseases from cardiovascular disease to cancer. The tools that we currently have to assess hormone health cannot make measurements non-invasively or in real-time. As a result, we not only lack understanding of how hormones affect us, but more importantly how we could rebalance them to improve quality of life and prevent disease.

Hormone profiling currently requires a blood test. Could this be done in real-time in an implantable or wearable device? We already have synthetic hormones but using these is plagued by side effects and health risks. Could we rethink design and delivery to overcome these debilitating effects? Implantable contraceptives are commonplace. Could we precisely modulate the hormone profile with controlled capture and release to rebalance the human system?

– Gemma Bale + Sarah Bohndiek, co-Programme Directors for Forecasting Tipping Points

Humanity produces about 500m tons of plastic per year. Although this is a very small fraction of total anthropogenic mass, plastic waste is an insidious problem given plastic’s heterogeneous distribution and use – likely trillions of discrete pieces, much of which are created for single-use applications. While concrete and steel tend to remain in places where humans need them, plastics end up dispersed around the entire planet, from the highest mountain peaks to the deepest ocean trenches. We know that waste plastics damage natural ecosystems and as they break down, the micro- and nanoplastics released travel vast distances in the air and waterways. Though the health effects are unclear, humans are ingesting significant amounts of microplastics with increasing links to oxidative stress, inflammatory response, alteration of lipid metabolism and beyond.

What would this look like? There’s a lot of research on more sustainable or biodegradable plastics, but adoption is limited by expense, scale, and an enormous range of material properties that must be reproduced in order to replace fossil-fuel derived formulations. Is there more we can do? Should we more definitively elucidate the scale of the problem? There are many examples in human history where we did not understand how bad something was for us – lead in pipes, paint, and petrol; arsenic and radium in consumer products; asbestos in construction. Can we lead a serious effort to more definitively establish the effects, if any, of microplastics on human and other health? How would attitudes change if we established that microplastics are toxic?

– Jenny Read, Programme Director for Robot Dexterity

A new wave of technologies is poised to redefine the computing landscape. They imply a future where the silos between different ecosystems are collapsed: where developers use code libraries from multiple languages in a single codebase and hardware communicates seamlessly whilst still being securely contained. This vision of the future is one of unprecedented security, interoperability, and composability.

We are beginning to catch glimpses of this future. WebAssembly (Wasm) is empowering developers to build high-performance applications in any major language and then execute the same code across any device, whether in the browser, cloud, locally, or at the edge. Developers may no longer need to target specific operating systems or hardware platforms, increasing efficiency and opening new markets. Hardware open standards like RISC-V and CHERI are coming of age. Low-cost edge computing has bloomed with platforms such as Raspberry Pi and Arduino. There’s increased adoption of memory-safe languages such as Rust, and experimentation with new microkernel operating systems.

We suspect this area is ripe for an explosion of innovation – and the implications are vast. Could this confluence redefine the role of the operating system, making existing ones obsolete? Could it bring an end to the ‘walled gardens’ of trillion-dollar app stores? Could entirely new industries emerge?

– Isabel Thompson, Product Lead

The world around us is changing at a phenomenal pace, but we are failing to prepare the students of today for the workplace of tomorrow. We still educate in a classroom, largely using pen and paper, asking children to solve problems that are graded as right or wrong based on a standard set of answers.

How will the classroom look in 2050?  Artificial intelligence is already transforming education, but it feels like we have only scratched the surface of how we educate our children in the era of large language models. What more is possible? Can neurotechnology transform our understanding of subconscious learning to maximise the value of artificial intelligence? What radically new hardware and software paradigms are needed to provide far more productive and natural learning environments?

– Gemma Bale + Sarah Bohndiek, co-Programme Directors for Forecasting Tipping Points

You might not have heard of Elysia timida, a small marine slug that – shy as it may be – shamelessly steals chloroplasts from its main food source, microscopic algae, to acquire the ability to photosynthesise. Imagine a world where farm herbivores, like sheep or cows, have been engineered to retain the chloroplasts from the grass they eat, such that a significant amount of their caloric intake comes from free and immensely abundant sunlight.

This could drastically change the unit economics of farming, reduce reliance on land for grazing animals, and even significantly change the environmental footprint of beef and lamb, two of the main contributors to agricultural CO2 and methane emissions. Where do we start with creating the mechanisms to model and test this? How do we appropriately benchmark the social and ethical considerations?

– Fabrizio Ticchiarelli-Marjot, Technical Specialist for Synthetic Plants

Tools like DNA sequencers and Cryo-EM have revolutionised our understanding of biology and are now being paired with large-scale AI algorithms to compound this impact. But most of the attention is going to the algorithms, with considerably less paid to the tooling. Meanwhile, most of the information gleaned from these tools comes from relatively static electromagnetic imaging. But LIFE IS NOT STATIC.

Are there tools that will allow us to better ‘see’ the temporal dynamics of our biological environments? Can we better visualise chemical signalling, metabolic pathways, structural mechanics? GraphCast recently demonstrated AI’s ability to forecast the weather with unprecedented accuracy leveraging years of data on weather dynamics – what new datasets do we need to unlock similarly powerful predictions in biology and health?

– Suraj Bramhavar, Programme Director for Nature Computes Better

Your muscles ache. You climb the stairs and you can somehow feel your lungs in your body. Your fingers go numb unexpectedly. You wake up in a daze and can’t pull yourself out of bed. Your heart rate shoots up to 180 bpm for no reason. You’re clearly sick, but modern medicine has little clue as to what’s wrong with you. 

Long COVID has raised the profile of health conditions that evade traditional diagnosis and treatments, but the integrated social and economic impact of these disorders remains wildly under-appreciated. With wide heterogeneity of symptoms across time and physiological systems, these disorders fall between the cracks of traditional specialties. It’s likely that AI combined with new sensing modalities can help, sussing out the systemic correlates we’re missing or not even bothering to look for.

Meanwhile, what excites us most about Scalable Neural Interfaces is the prospect of developing a single technological insertion point for diagnosing and addressing systemic disorders. What other physiological levers should we be trying to engineer in similar ways? The microbiome has garnered quite a bit of attention, but how about the immune, vascular, mitochondrial, epigenetic, metabolic, or endocrine systems?

– Ilan Gur, CEO

What if you could reprogramme cells to wirelessly communicate over centimetre-scale distances? While many endogenous signalling mechanisms have already been discovered and programmed with the help of synthetic biology, emerging work in fields such as magnetogenetics and sonogenetics may vastly expand the spatiotemporal scale of cell-to-cell communication. For example, mammalian cells can be genetically reprogrammed to express hollow gas vesicle proteins which could act as cellular-scale ultrasound ‘emitters’. These emitters could then activate mechanosensitive ion channels in different cells to act as a cellular-scale ultrasound ‘receiver’. Moreover, by coupling breakthroughs in engineered biology with advances in low-power silicon CMOS, where tens of thousands of transistors can fit in the surface area spanned by a single neuron, we may be able to build cellular scale bio-electronic hybrids which fundamentally change understanding of what it means to be a multicellular organism.

This thus raises the question: if these new signalling mechanisms could be hacked, what new functionalities could we enable within humans? Could we wirelessly bypass damaged neural circuitry for patients with spinal cord injury? Could we programme wireless networks of engineered cells to sense and destroy cancerous cells? Could we build distributed biological computing nodes to enable superhuman reasoning? Could we have non-invasive brain-computer interfaces purely based on endogenous signalling from neurons?

– Jacques Carolan, Programme Director for Precision Neurotechnologies

Plants are exposed: not just to the elements, but also to a variety of devastating pests and pathogens. For example, cereal rust, which causes losses up to $5bn around the world each year; ash dieback, which is forecasted to decimate UK ash tree populations by up to 80%; and late blight, which destroys up to 30% of potato yield across the globe. We vaccinate ourselves, our livestock, and our pets against common diseases. Can we develop vaccines for plants, priming an immune response to prevent the initiation and spread of plant disease?

While animal immunity adapts and learns from exposure to pathogens, plants have far less flexible innate immune systems. Chemical priming provides one avenue to narrowly boost immune response, but ideally plants would actively develop immunity to new and unexpected pathogens. This could even be triggered by pre-emptive signalling from infected neighbours. What would be the mechanism of action of such a vaccine? How could it be administered to allow for uptake from every plant in a field? Without a circulatory system, how would a plant ensure the vaccine and immune response make it to the site of an attack?

– Angie Burnett, Programme Director for Synthetic Plants

Arguably one of the biggest dangers from ‘AI’ comes from recommendation algorithms and their propensity to favour content distribution generating the most clicks (often pushing our information diet towards the extreme). An entire economy has been built around this engagement model, and its implications for society are profound. Is there a technology solution to this dilemma? Can we create a safeguarded AI tuned to our values? Can neurotechnologies serve as attention bandpass filters? What else?

– Suraj Bramhavar, Programme Director for Nature Computes Better

Cardiovascular surgery has been transformed by technological innovation through the advent of interventional cardiology. In this field, complex procedures are performed entirely through blood vessels using cardiovascular stents, reducing surgery time, minimising complications, and enabling care at scale compared with open-heart surgery. Can we use the same approach to transform treatment of brain disorders— the leading cause of disability worldwide?

Research has shown that neuroendovascular stents can be augmented to record low resolution neural signals, but there has otherwise been very little exploration of vascular neurotech. What advances in materials science would enable neuroendovascular access to the entire volume of the human brain? What entirely new platforms can we use to record and modulate biologically relevant signals solely through blood vessels? Can we leverage cutting-edge advances in CMOS manufacturing to achieve I/O data rates in vascular interfaces comparable to those of implanted BCIs?

– Jacques Carolan, Programme Director for Precision Neurotechnologies

The basic premise of how we obtain our food has not changed since the first agricultural revolution 12,000 years ago: the plan is still essentially to plant a seed and hope for the best. Almost every food that we eat is produced outside, at the mercy of the weather, with us passively waiting and praying for clement conditions. The risks of this strategy for food security and abundance are obvious, and there are any number of lessons from history as to what happens when harvests fail.

If we are to become a multiplanetary species, or even to continue living on Earth with a growing population without irreparable damage to the environment, then we need new ways to feed ourselves. We will eventually need to develop the ability to grow food indoors, and to move beyond using just incident sunlight to power production of the global food supply. Could we use nuclear power to grow food underground and/or on Mars without having to bother terraforming the whole planet? Could we work out how to grow an apple without growing the rest of the tree? In doing so, could we grow them faster? What is a reasonable rate of production for apples anyway: one per month? Or one per week, or per day, or per hour?

– Mark Symes, Programme Director for Exploring Climate Cooling

This phrase gained popularity at Sun Microsystems during the rise of the modern computing era. As more computing power becomes harder and harder to realise, it has taken on new meaning. Currently, the energy requirements of sending information dramatically exceed those of processing information (by a factor of ~10,000). This represents the principle bottleneck of modern computing systems.

While long-distance communication systems have adopted complex protocols and error correction techniques to transmit bits at maximal efficiency (approaching the Shannon limit), today’s computing systems still rely on an incredibly primitive form of information transfer (on-off keying). Is there an opportunity to rethink how we transfer information between or within chips, to approach the Shannon limit, and to fundamentally rethink existing computer architectures?

– Suraj Bramhavar, Programme Director for Nature Computes Better

Diagnostic tools for the early detection of infection are laborious and costly when deployed at scale, as evidenced by the results of global efforts to contain COVID-19 outbreaks through testing programmes that used genetic or biochemical analytical methods. What if we could use plants that can detect the presence of common human and animal pathogens? 

The use of sentinel plants has long been employed in integrated pest management systems or strategies to detect environmental pollutants, and a similar rationale can be leveraged for early detection and mapping of pathogens outbreaks. Plants are able to produce a variety of signals in response to cellular events – these include the release of intracellular compounds, electrical signals, exuded compounds in the rhizosphere, and gaseous hormones such as ethylene. Any of these signals could be engineered to be released in response to the presence of any microorganisms of choice.

These analogue signals could be employed to directly connect a global network of inexpensive, self-replicating, adaptable and easily deployable sentinel plants to epidemic/pandemic predictive models – which would ensure accurate, granular and timely data on the spread of pathogens, saving countless lives and reducing the impact of such event on global health systems and our personal freedom.

– Fabrizio Ticchiarelli-Marjot, Technical Specialist for Synthetic Plants