< Previous3 /9 2 /10 5 TRL /9 D IFFRACTIVE NEURAL NETWORKS, which are in the early Prototype Stage, is the field of research concerned with creating the world’s first physical passive neural networks by 3D Printing them, rather than programming them, and using light waves, not electrons, to perform machine learning tasks, such as image analysis, feature detection and object classification, at the speed of light without the need to rely on any external compute or power source. Recently there have been a couple of interesting breakthroughs in the space, in the automated production of these types of neural networks, and their low cost, and ease of deployment, which makes them potentially a very interesting twist on a popular technology. DEFINITION Diffractive Neural Networks is a form of physical Artificial Intelligence that is printed and encoded into physical objects rather than being manifested in machine code. EXAMPLE USE CASES Today the first prototype Diffractive Neural Networks are being used in image detection, image analysis, and object classification to test the theory and refine the technology. In the future the primary use case of the technology will be passive neural network applications where speed is useful or important. FUTURE TRAJECTORY AND REPLACABILITY Over the next decade interest in the field will continue to accelerate, and interest and investment will continue to grow, albeit from a very low base, primarily led by university grants. In time the technology will continue to be refined and proven with researchers looking into new ways to produce and manufacture these kinds of networks automatically and at speed. While Diffractive Neural Networks are in the early Prototype Stage, over the long term it will be enhanced by advances in 3D Printing and Nano-Manufacturing, but at this point in time it is not clear what it will be replaced by. MATTHEW’S RECOMMENDATION In the short to medium term I suggest companies put the technology on their radars, and re-visit it every few years until progress in the space accelerates. 15 SECOND SUMMARY Accessibility Affordability Competition Demonstration Desirability Investment Regulation Viability 3 6 5 4 5 2 1 8 2010 2014 2017 2028 2038 STATUS PRIMARY GLOBAL DEVELOPMENT AREAS IMPACT DIFFRACTIVE NEURAL NETWORKS STARBURST APPEARANCES: ‘19, ‘20, ‘21, ‘22, ‘23, ‘24 EXPLORE MORE. Click or scan me to learn more about this emerging tech. 320311institute.com MRL3 /9 9 /10 4 TRL /9 D NA NEURAL NETWORKS, which is in the early Prototype Stage, is the field of research concerned with creating a new so called “Wet Artificial Intelligence” technologies using nothing more than biological components, in the first case, DNA, to create complex neural networks that one day could be integrated with and programmed into molecular machines, and even the molecular machinery of the human body, in essence, helping turn the human body into a biological supercomputer. DEFINITION DNA Neural Networks is a form of Artificial Intelligence, also known as Wet AI, that is built from DNA rather than machine code. EXAMPLE USE CASES Today the first DNA Neural Network prototypes are being used to train the networks to identify handwriting before being refined. In the future the primary applications of the technology will be to bring the power of Artificial Intelligence to new environments, such as liquids, where their use cases will be as numerous as their digital counterparts. FUTURE TRAJECTORY AND REPLACABILITY Over the next decade interest in the field will continue to accelerate, and interest and investment will continue to grow albeit from a very low base, primarily led by organisations in the healthcare sector, with support from government funding, and university grants. In time we will see researchers become increasingly capable at building and deploying increasingly complex DNA Neural Networks that have a wide variety of applications, but it is also likely that productising the technology will be hampered by regulation. While DNA Neural Networks are in the early Prototype Stage, over the long term they will be enhanced by advances in 3D Bio-Printing, Biological Computing, Bio-Manufacturing, CRISPR Gene Editing, DNA Computing, DNA Nanoscience, DNA Synthesis, Nano-Machines, Nano-Manufacturing, and Semi-Synthetic Cells, but at this point in time it is not clear what they will be replaced by. MATTHEW’S RECOMMENDATION In the short to medium term I suggest companies put the technology on their radars, and re-visit it every few years until progress in the space accelerates. 15 SECOND SUMMARY Accessibility Affordability Competition Demonstration Desirability Investment Regulation Viability 2 2 4 3 8 2 1 7 1997 2009 2016 2032 2046 STATUS PRIMARY GLOBAL DEVELOPMENT AREAS IMPACT STARBURST APPEARANCES: ‘19, ‘20, ‘21, ‘22, ‘23, ‘24 DNA NEURAL NETWORKS EXPLORE MORE. Click or scan me to learn more about this emerging tech. 321311institute.com MRL3 /9 10 5 TRL /9 EVOLUTIONARY ARTIFICIAL INTELLIGENCE, which is in the Prototype Stage, is the field of research concerned with developing AI’s that are capable of evolving their own code in much the same way that animals combine together different DNA strands to create new lifeforms and evolutionary alternatives. While this is a highly promising technology that could yield a great number of interesting and surprising results it has to be said it’s also highly likely that it will produce all manner of unpredictable outcomes none of which might be controllable or favourable. Recently there have been a number of breakthroughs in this field from companies including Google who have created AI’s capable of mutating and evolving their “programming and code” autonomously. DEFINITION Evolutionary Artificial Intelligence are unsupervised neural networks that rely on fitness functions rather than training to improve and solve problems. EXAMPLE USE CASES With examples such as AutoML-Zero which resembles natural evolution with the AI code improving every generation with little or no human interaction by evaluating the top performing candidate algorithms for a particular task and then mathematically combining them together and “evolving” them this is an interesting field with many applications including the development of better Generalised Robots and superior AI’s - especially if it’s combined with Open Ended AI systems. FUTURE TRAJECTORY AND REPLACABILITY Over the next decade interest in the field will continue to accelerate, and interest and investment will continue to grow at an accelerating rate, primarily led by organisations in the technology sector, with support from government funding and university grants. In time we will see Evolutionary AI become common place and see evolution taking place at digital speed at which point it’s highly unlikely that these systems or their outcomes will be fully controllable. Furthermore, as AI gets better at designing and generating its own algorithms this is literally a technology that could take on a life all of its own which is both fascinating and dangerous. While Evolutionary AI is in the Prototype Stage, over the long term it will be enhanced by advances in Artificial Intelligence, Explainable AI, Open Ended AI, Quantum Computing, and many other complimentary fields, but at this point in time it is not clear what it will be replaced by. MATTHEW’S RECOMMENDATION In the short to medium term I suggest companies put the technology on their radars, explore the field, establish a point of view, experiment with it, and forecast out the implications of the technology. 15 SECOND SUMMARY Accessibility Affordability Competition Demonstration Desirability Investment Regulation Viability 3 3 2 7 9 6 2 9 1969 1975 2018 2021 2032 STATUS PRIMARY GLOBAL DEVELOPMENT AREAS IMPACT EVOLUTIONARY AI STARBURST APPEARANCES: ‘22, ‘23, ‘24 322311institute.com EXPLORE MORE. Click or scan me to learn more about this emerging tech. MRL5 /9 9 /10 7 TRL /9 E XPLAINABLE AI (XAI), which is in the early Productisation Stage, is the field of research concerned with developing new AI tools and techniques that enable AI’s to explain their decision making processes to humans using natural language, visualisations, or in other formats. Recently there have been a variety of breakthroughs in the field when it comes to “reading the mind” of so called neural Black Box AI’s which include the ability to visualise decision trees, as well as use natural language to ask AI’s clarifying questions, but despite these developments there is still a long way to go before these systems are robust and trustworthy enough to be incorporated into high risk and high stakes applications. DEFINITION Explainable AI is a set of tools and techniques that allow humans to comprehend, query, and thereby trust the outputs of AIs. EXAMPLE USE CASES As AI becomes more embedded into the digital fabric of society there are an increasing number of examples and reasons why it needs to be able to explain its actions and decisions, such as in the defense, healthcare, financial, legal, and transportation sectors especially, whether it’s military personnel understanding AI target selections or doctors understanding the reasoning behind suggesting specific medical treatments, and many other examples. FUTURE TRAJECTORY AND REPLACABILITY Over the next decade interest in the field will continue to accelerate, and interest and investment will continue to grow at an accelerating rate, primarily led by organisations in the technology sector, with support from government funding and university grants. In time we will see the vast majority of both enterprise and consumer, private and public, AI’s become embodied with this capability but with Explainable AI development and investment massively lagging behind conventional AI development it will be a while before researchers develop robust models and incorporate them into the majority of AI models. While Explainable AI is in the early Productisation Stage at this point in time it is not clear what it will be replaced by. MATTHEW’S RECOMMENDATION In the short to medium term I suggest companies put the technology on their radars, explore the field, establish a point of view, experiment with it, and forecast out the implications of the technology. 15 SECOND SUMMARY Accessibility Affordability Competition Demonstration Desirability Investment Regulation Viability 4 3 3 6 9 4 2 8 1993 2014 2018 2022 2034 STATUS PRIMARY GLOBAL DEVELOPMENT AREAS IMPACT EXPLAINABLE AI STARBURST APPEARANCES: ‘22, ‘23, ‘24 323311institute.com EXPLORE MORE. Click or scan me to learn more about this emerging tech. MRL8 /9 7 /10 9 TRL /9 F EDERATED ARTIFICIAL INTELLIGENCE, which is in the Prototype Stage and early Productisation Stage, is the field of research concerned with finding new ways to train and develop Artificial Intelligence models without the need to rely on capturing and transporting vast volumes of data back to centralised cloud datacenters for processing, instead pushing those tasks to the devices at the edges of the network, which has the added benefit of not compromising user privacy, dramatically reducing network latency, and creating smarter models that consumers can leverage immediately. Currently one of the biggest issues that organisations developing Artificial Intelligence platforms have is capturing enough training data to train their models, and as a result this technology is potentially invaluable. DEFINITION Federated Artificial Intelligence allows disparate devices to collaboratively learn a shared prediction model while keeping all the training data on device, decoupling the need to store data in centralised data centers. EXAMPLE USE CASES Today we are using Federated Artificial Intelligence to learn about, and then improve, the usability of smartphones, and messaging systems. In the future the primary use case for this technology will be to use it to tap into the data and powerful devices at the edge of the network to create smarter models. FUTURE TRAJECTORY AND REPLACABILITY Over the next decade interest in the field will continue to accelerate, and interest and investment will continue to grow at an accelerating rate, primarily led by organisations in the Technology sector. In time we will see this technology become one of the primary methods organisations use to train their models, and as the devices at the edge become more capable, powerful, and smart, whether those devices are drones and robots, smartphones or vehicles, and everything and anything in between, it is inevitable that they will assume more of the training workload. While Federated Artificial intelligence is in the Prototype Stage and early Productisation Stage, over the long term it will be enhanced by advances in Artificial Narrow Intelligence, Artificial General Intelligence, Distributed Computing, Neural Processing Units, and Sensor Technology, but at this point in time it is not clear what it will be replaced by. MATTHEW’S RECOMMENDATION In the short to medium term I suggest companies put the technology on their radars, explore the field, establish a point of view, experiment with it, with a view to implementing it, and forecast out the potential implications of the technology. 15 SECOND SUMMARY Accessibility Affordability Competition Demonstration Desirability Investment Regulation Viability 7 7 6 7 8 4 2 8 2006 2011 2016 2018 2030 STATUS PRIMARY GLOBAL DEVELOPMENT AREAS IMPACT FEDERATED ARTIFICIAL INTELLIGENCE STARBURST APPEARANCES: ‘19 EXPLORE MORE. Click or scan me to learn more about this emerging tech. 324311institute.com MRL3 /9 9 /10 4 TRL /9 L IQUID ARTIFICIAL INTELLIGENCE, a General Purpose Technology, which is in the Prototype Stage, is the field of research concerned with trying to develop liquid neural networks, which work in a similar way to the human brain, that can match or exceed the efficiency and performance of their more traditional Artificial Narrow Intelligence (ANI) counterparts. While this technology is very young recent breakthroughs include the development of the first Water- Ion based liquid AI’s which were able to perform simple calculations and computations. As research across multiple complimentary “aqueous” technology fields progresses this will be an interesting technology to watch, but that said it has a significant way to go before it’s productised. DEFINITION Liquid Artificial Intelligence is the use of aqueous liquids and their properties to perform neural network-like calculations and computations. EXAMPLE USE CASES While the future use cases of this technology could be diverse, from using it to perform healthcare related compute tasks within the human body, as well as any other kind of compute tasks in any manner of suitable aqueous solutions, its sweet spot is as yet are unclear. Given the trajectory of its development though and its similarity to biological intelligence there is a high likelihood that it could have extreme utility. FUTURE TRAJECTORY AND REPLACABILITY Over the next decade we will continue to see interest and investment in this space accelerate, predominantly led by government grants. While the usefulness of this technology is without question the pace and trajectory of its future development will rely heavily on developments in other complimentary technology fields which could help or hinder its progress. As such it has a fuzzy future but an interesting one nonetheless. While Liquid Artificial Intelligence is still in the Prototype Stage it could be enhanced by advances in other aqueous technology fields including Liquid Computing and Electronics, as well as 3D Bio-Printing, Organoids, Nanotechnology, Quantum Computing, Synthetic Biological Intelligence, and other technologies, however over the long term it’s unclear what it could be superseded by. MATTHEW’S RECOMMENDATION In the short to medium term I suggest companies put the technology on their radars, and re-visit it every few years until progress in the space accelerates. 15 SECOND SUMMARY Accessibility Affordability Competition Demonstration Desirability Investment Regulation Viability 2 4 4 8 9 2 2 9 1962 1978 2023 2045 2065 STATUS PRIMARY GLOBAL DEVELOPMENT AREAS IMPACT LIQUID ARTIFICIAL INTELLIGENCE STARBURST APPEARANCES: ‘24 325311institute.com MRL EXPLORE MORE. Click or scan me to learn more about this emerging tech.9 /9 8 /10 9 TRL /9 M ACHINE VISION, which is in the Mass Adoption Stage, is the area of research concerned with developing the systems and tools that allow machines to see and understand the world around them. Recently there have been numerous breakthroughs in the field thanks to dramatic advances in Artificial Intelligence, which now means that machines are increasingly adept at understanding, sensing, and interacting with the world around them. Whether it’s autonomous vehicles, security, or robotics, and many other applications besides, arguably developing more advanced AI models has been the breakthrough the technology needed in order to really come to life and live up to its promise of not just helping machines see the world around them, but also interact with it in new and bold ways. DEFINITION Machine Vision harnesses Deep Learning algorithms to automatically analyse, interpret and inspect still images and streaming video. EXAMPLE USE CASES Today we are using Machine Vision across a wide range of areas, from using smartphones to diagnose cancers, and create better manufacturing and warehouse robots, to safer autonomous vehicles, and more capable surveillance systems capable of detecting criminal intent, and many more. In the future the primary use case of the technology will be as it is today, giving machines the ability to see, interpret and interact with the world around them in improved ways. FUTURE TRAJECTORY AND REPLACABILITY Over the next decade interest in the field will continue to accelerate, and interest and investment will continue to grow at an accelerating rate, primarily led by organisations in the Defence and Technology sector, with support from government funding, and university grants. In time we will see the technology evolve to include the analysis of the entire electromagnetic spectrum, and see it combined with other sensing technologies that provide intelligent machines with even deeper insights into the world around them. While Machine Vision is in the Mass Adoption Stage, over the long term it will be enhanced by advances inArtificial General Intelligence, CRISPR Gene Editing, Diffractive Neural Networks, Artificial Narrow Intelligence, Lensless Cameras, Optics, and Sensor technology, but at this point in time it is not clear what it will be replaced by. MATTHEW’S RECOMMENDATION In the short to medium term I suggest companies put the technology on their radars, explore the field, establish a point of view, experiment with it, with a view to implementing it, and forecast out the potential implications of the technology. 15 SECOND SUMMARY Accessibility Affordability Competition Demonstration Desirability Investment Regulation Viability 9 9 5 9 9 8 5 9 1965 1978 1983 1988 2023 STATUS PRIMARY GLOBAL DEVELOPMENT AREAS IMPACT STARBURST APPEARANCES: ‘17, ‘18, ‘19, ‘20, ‘21, ‘22, ‘23, ‘24 MACHINE VISION EXPLORE MORE. Click or scan me to learn more about this emerging tech. 326311institute.com MRL3 /9 6 /10 4 TRL /9 M ECHANICAL ARTIFICIAL INTELLIGENCE, which is in the Prototype Stage, is the field of research concerned with developing physical forms of Artificial Intelligence (AI) that can be embedded into materials and objects to give them AI-like adaptable, reactive, and tunable properties. Recently there have been a few breakthroughs in the field including the development of a solid “Architected Material” by an algorithm whose entire structure resembles that of a neural network that can adapt its physical and mechanical properties according to the environment it’s in and the stimulii it’s subjected to. While this field is still young it potentially opens the door to an exciting class of new materials that could also compliment other classes of materials including Metamaterials and Reactive Materials. DEFINITION Mechanical Artificial Intelligence is the embodyment of neural network-like technologies and schema within the physical structures of materials and objects. EXAMPLE USE CASES Today the primary use cases for Mechanical Artificial Intelligence is in materials that have to adapt and react to different loads and stresses on the fly, such as aircraft airframes, but over the longer term it could also be used to develop new shape shifting materials and possibly smart materials. FUTURE TRAJECTORY AND REPLACABILITY Over the next decade we will continue to see interest in this field accelerate, albeit from a very low base, led by government grants. While the concept of Mechanical Artificial Intelligence is still young as we continue to see exponential advances across different technology categories this could be a dark horse technology to watch. It’s potential could also be multiplied when combined with other technological advancements. While Mechanical Artificial Intelligence is still in the early Prototype Stage over the longer term it could be enhanced by advances in AI including 3D and 4D Printing, Diffractive Neural Networks and DNA Neural Networks, Electronics, Energy, Materials, Metamaterials, Quantum Computing, Robotics, Sensors, and other technologies, however over the long term it’s unclear what it could be superseded by. MATTHEW’S RECOMMENDATION In the short to medium term I suggest companies put the technology on their radars, establish a point of view, and re- visit it every few years until progress in the space accelerates. 15 SECOND SUMMARY Accessibility Affordability Competition Demonstration Desirability Investment Regulation Viability 3 5 5 5 7 2 2 7 1981 1989 2022 2042 2064 STATUS PRIMARY GLOBAL DEVELOPMENT AREAS IMPACT MECHANICAL ARTIFICIAL INTELLIGENCE STARBURST APPEARANCES: ‘23 EXPLORE MORE. Click or scan me to learn more about this emerging tech. 327311institute.com MRL2 /9 10 4 TRL /9 M ETA ARTIFICIAL INTELLIGENCE, which is in the Prototype Stage, is the field of research concerned with creating AI’s that are capable of creating computers and computer-like resources that run within the same AI in simulation which themselves can then, in turn, run other AI’s and code - thereby in effect creating a “multi- dimensional” software construct where, provided there is enough real world compute resource one single AI can run and “play host to” an infinite number of other AI’s and compute-like systems within itself. Not only is this technology revolutionary but as AI’s get better at designing and evolving their own code and programming it could spawn all manner of new compute and intelligence paradigms with almost limitless possibilities that today are difficult to comprehend. DEFINITION Meta Artificial Intelligence is the practise of creating computers and compute-like systems within virtual neural networks that can run code. EXAMPLE USE CASES Today this is a very nascent field but so far the AI’s that have been able to create computers within themselves that have then, in turn, been able to run other AI’s and code, have been used to split up incredibly complicated tasks, spread the workload among different virtual “AI Cores” and then re-combine them. FUTURE TRAJECTORY AND REPLACABILITY Over the next decade interest in the field will continue to accelerate, primarily led by organisations in the technology sector, albeit from a very low base, with limited support from government funding and university grants. In time we could see Meta Artificial Intelligence become ubiquitous in much the same way that today “dumb” virtual servers that run on physical servers are ubiquitous in today’s modern datacenters, except in this case the systems won’t just be dumb they will be intelligent, self-designing, self-evolving, and self-improving at digital speed. While Meta Artificial Intelligence is in the Prototype Stage, over the long term it will be enhanced by advances in Artificial Intelligence, Meta Computing, and Neuromorphic Computing, but at this point in time it is not clear what it will be replaced by. MATTHEW’S RECOMMENDATION In the short to medium term I suggest companies put the technology on their radars, explore the field, establish a point of view, experiment with it where possible, and forecast out the implications of the technology. 15 SECOND SUMMARY Accessibility Affordability Competition Demonstration Desirability Investment Regulation Viability 2 2 2 7 8 2 1 8 2004 2018 2022 2038 2045 STATUS PRIMARY GLOBAL DEVELOPMENT AREAS IMPACT META ARTIFICIAL INTELLIGENCE STARBURST APPEARANCES: ‘23, ‘24 328311institute.com EXPLORE MORE. Click or scan me to learn more about this emerging tech. MRL9 /9 7 /10 9 TRL /9 N ATURAL LANGUAGE PROCESSING, a GENERAL PURPOSE TECHNOLOGY, which is in the Mass Adoption Stage, is the field of research concerned with helping machines analyse and synthesise natural language, whether that language is in speech or written form. Recently there have been significant breakthroughs in the technology, including the ability for machines to translate between hundreds of different languages on the fly, as well as their ability to understand subtle variations in context and tone, as well as their ability to synthesise speech at such a high level it fools humans. DEFINITION Natural Language Processing is the application of computational techniques to the analysis and synthesis of natural language and speech. EXAMPLE USE CASES Today we are using Natural Language Processing in a number of ways including behavioural computing, natural language translation, speech to text and vice versa, and semantic analysis of literary works. In the future the primary uses of the technology will include breaking down translation barriers, enabling frictionless human-machine communication, and using AI to analyse and unlock the information contained within text and voice based content. FUTURE TRAJECTORY AND REPLACABILITY Over the next decade interest in the field will continue to accelerate, and interest and investment will continue to grow at an accelerating rate, primarily led by organisations in the Defence and Technology sector, with support from government funding. In time we will see machines become increasingly adept at understanding natural human language, with their accuracy edging towards 100 percent in all fields, and they will become increasingly adept at communication with us in natural language that is imperceivable from a real person. While Natural Language Processing is in the Mass Adoption Stage, over the long term it will be enhanced by advances in Artificial Intelligence, Behavioural Computing, Federated Artificial Intelligence, and Intelligence Processing Units, but at this point it is not clear what it will be replaced by. MATTHEW’S RECOMMENDATION In the short to medium term I suggest companies put the technology on their radars, explore the field, establish a point of view, experiment with it, with a view to implementing it, and forecast out the potential implications of the technology. 15 SECOND SUMMARY Accessibility Affordability Competition Demonstration Desirability Investment Regulation Viability 9 7 4 9 9 8 5 9 1961 1964 1969 1985 2028 STATUS PRIMARY GLOBAL DEVELOPMENT AREAS IMPACT NATURAL LANGUAGE PROCESSING STARBURST APPEARANCES: ‘17, ‘18, ‘19 EXPLORE MORE. Click or scan me to learn more about this emerging tech. 329311institute.com MRLNext >