What is ARIA looking to support?

The Advanced Research and Invention Agency or ARIA is set to be the Government’s newest funding agency. ARIA is designed to fund projects with potential to produce transformative technological change, or a paradigm-shift in an area of science. This is achieved through a series of research programmes. 

In our last blog we introduced ARIA and the proposed funding types on offer. In this blog we delve into the first series of programmes and opportunities and also take a look at what might  be coming up. 

Over the last two years the UK Government has been busy recruiting Programme Managers. These Programme Managers have freedom and control to dynamically channel funding, shift project objectives and milestones, and manage risks, to keep their overall 3-5 year research programme focussed around a coherent but evolving vision. 

The Programmes  

Programmable plants: a technology platform for sustainable abundance 

Plants have paved the way for human existence and hold tremendous potential to solve some of our most pressing challenges such as food insecurity, climate change and environmental degradation. Programmable plants can secure our future on earth – providing not just food, but a sustainable and thriving biosphere for future generations. 

Programme Director, Angie Burnett, plans to develop a detailed thesis surrounding the core beliefs that a paradigm shift is needed to accelerate innovations to address the coupled challenges of sustainable food supply and a stable climate, that plants offers massive potential for low-cost, sustainable resources, and that we can predictably and efficiently develop novel plants via gene editing and genetic modification. 

Precisely interfacing with the human brain at scale 

Neurological and neuropsychiatric disorders are the cause of an overwhelming societal and economic burden. It is becoming increasingly evident that targeted interaction with the human nervous system can improve the human condition across an incredibly wide range of disease states and cognitive domains. 

Programme Director, Jacques Carolan, aims to support researchers developing tools to interface with the human brain with unprecedented precision. This programme will unlock new therapeutic methods to understand, identify and treat neurological and neuropsychiatric disorders. 

This programme aims to develop new tools that can interact with the central nervous system at the circuit level to understand the onset and progression of disease, identify biomarkers of disease states and ultimately to treat these disorders. 

Funding targets include a variety of institutions (e.g. academic research groups, start-ups, established industry) across the broad technical areas: 

• Development of next-generation precision neurotechnologies. Teams should focus on  circuit-level access to the brain, with cell type specificity and across distributed macro- and micro-brain circuits, targeting in vivo portable proof of concept within four years. 
• Applying precision neurotechnologies to demonstrate the controllable transition between brain states, in vivo. This is highly likely to involve interdisciplinary teams. 

Scoping our planet: a new lens on climate science 

Programme Directors Gemma Bale and Sarah Bohndiek, are looking to address current Earth system measurements which have serious gaps that lead to uncertainties in weather forecasting and climate predictions. By harnessing the power of novel optics and photonics we can fill these gaps, equipping society to respond confidently to the climate crisis. 

Nature computes better – let’s catch up 

Programme Director, Suraj Bramhavar, is addressing how to redefine the way computers process information by exploiting principles found ubiquitously in nature. In doing so, we can better understand how the natural world around us performs computation and build dramatically more efficient computers.  

This programme aims to identify defined workloads through which to focus development, for example (but not exhaustive), matrix multiplication, matrix inversion, and monte carlo sampling. Regardless of the kernel chosen, well-defined sets of benchmarks will be required to evaluate performance. Teams would be charged with showing that, for a set of target workloads, specific performance/cost metrics can be met representing a radical step-change over the projected state-of-the-art. Funding will also include areas such as hardware for AI training. AI research is increasingly governed by the efficiency with which practitioners can access vast compute resources to train new models, with workloads targeted limited by the hardware available. 

Mathematics and modelling are the keys we need to unlock safe and transformative AI 

Programme Director, David ‘davidad’ Dalrymple, envisions a pathway to leverage frontier AI itself to collaborate with humans to construct a “gatekeeper”: a targeted AI whose job is to fully understand the real-world interactions and consequences of an autonomous AI agent, and to ensure the agent only operates within agreed-upon safety guardrails and specifications for a given application. We don’t yet have known technical solutions to ensure that powerful AI systems interact as intended with real-world systems and populations. A combination of scientific world-models and mathematical proofs may be the answer to ensuring AI provides transformational benefit without harm. 

A gatekeeper safeguard is anticipated to have three distinct components: 

1. An explainable, auditable scientific mathematical model of task-relevant aspects of the real world with defined quantified specifications of safety.

2. To prove probabilistic bounds and provide proof certificates

3. To deploy a deep reinforcement learning training loop to adapt the model(s) or neural networks with high verifiability. 

This programme expects to fund scaffolding (models, specifications, proofs) including theory, backend development and human-computer interfaces; machine learning including world-modelling, proof-searching, and certification; and applications to demonstrate that “gatekeeper AI” as a workflow can be used to create and maintain decision-support tools and/or safeguarded autonomous AI systems that deliver value in practice for specific tasks. 

Managing our climate and weather through responsible engineering 

Programme Director, Mark Symes focus is on exploring options for actively cooling the earth. As it stands if an abrupt alteration in a climate system were to unfold, we would have no tools to mitigate the effects.  

Through short-duration, small-scale and geographically-confined field trials, this programme aims to answer fundamental questions as to the practicality, measurability and controllability of such technologies. In answering these questions, we plan to fund not only the field trials themselves, but also the necessary modelling, simulation, indoor experiments, observation and monitoring required to support the trials, as well as research into the legal, ethical, governance and geopolitical dimensions of the approaches under investigation. 

We anticipate supporting research into approaches for reducing global temperatures across the full range of science and engineering disciplines. We also expect to support projects across the social sciences that are of direct relevance to those approaches (including, but by no means limited to, consideration of public perception, potential legal, ethical, regulatory and governance frameworks, ethics, community engagement, and the economic impact of those approaches). 

Smart machines need smarter bodies 

Throughout history, humans have used tools and machines to reduce the burden of physical labour. We are entering a new era with robots smart enough to act independently in complex and dynamic environments. But smart machines with dumb bodies will only get us so far. 

Within this opportunity Programme Director, Jenny Read, is exploring overcoming the primary bottleneck to robotics adoptions: manipulation. The focus is on improving robotic dexterity primarily through advances in hardware, ultimately to realise the Darwin paradigm of bio-inspired robotics. Although big tech companies are working on robots with unprecedented ability and incorporating recent advancements in reinforcement learning and multi-modal large language models their results will fundamentally be limited by hardware restrictions.  

Support in this programme is planned for development of new modes of sensing, transmission of sensory information, and actuation through hardware advances that benefit from co-design and integration with advanced software and controls. 

In the early stages of the programme, funding is anticipated to cover advances in individual components. In later stages, the plan is to combine advances made both within and beyond the programme to develop new manipulators, demonstrating a paradigm-shift in robotic abilities and establishing the basis for a powerful new industry that can help society better address the labour challenges of tomorrow. 

 How to get involved 

Head to the ARIA agency website to read through the detailed programme theses.  

To keep track of the open and upcoming opportunities you can head over the Funding Tracker on TBAT’s website, or talk with us directly to find out more. 

Book an appointment today