The Promise and Problems of Deep Tech: What is in Store for 2021
Industry: Built Environment (USS)
The COVID-19 pandemic will forever define 2020. But Deep Tech gave us the tools to respond swiftly, allowing us to enter 2021 with a glimmer of hope.
Just a year into the pandemic, vaccines are now increasingly being rolled out across the globe. Organisations are adapting to a low-contact, remote world. Supply chains are dealing with disruption; industries are preparing for a more liveable, sustainable world.
But as the spotlight shines brighter on the potential of Deep Tech to promote social and economic good, it also reveals the perils of such innovations.
For example, inaccurate or incomplete data can lead automated systems to make drastically wrong decisions. And for all the good AI brings, it can also endanger the planet if it continues to burn energy at its current pace.
As such issues come to the fore, we can expect scientists and entrepreneurs to devote more resources to coming up with solutions in 2021.
These Deep Tech promises and problems were discussed during The Promise of Deep Tech: The Year Ahead, a webinar presented by SGInnovate and A*StartCentral on 20 January. SGInnovate also held a related webinar, The Next Step for Deep Tech in Asia, on 13 January.
1. Digital biomarkers will enable precise diagnostics even with patients being miles away
Imagine swallowing a pill for regulating blood pressure and knowing its real-time effect on your body. A wearable sensor could track the pill’s immediate impact and even send this information to your doctor. This data, along with the method of detecting and describing it, is called a digital biomarker.
Karger, a publisher of journals in the Health Sciences, defines digital biomarkers as “objective, quantifiable physiological and behavioural data that are collected and measured by means of digital devices such as portables, wearables, implantables, or digestibles.”
Think, for example, of how a smartwatch can detect your blood sugar levels, blood pressure, and heart rate — and how these biomarkers differ from physical ones like saliva or blood.
Speech and language can also be sources of digital biomarkers. For instance, scientists are studying the use of vocal biomarkers to detect symptoms of depression, dementia, and Parkinson’s disease.
“It’s precise enough that clinicians can diagnose or prescribe based on those results,” said Cort Isernhagen, Senior Vice President & Managing Director, Lux Research, during The Promise of Deep Tech webinar.
It is not just a healthcare play, though. “It’s also, in my opinion, a very interesting consumer product play for the future,” added Isernhagen.
For instance, a device might track the effect of packaged food on your gut health as you consume it. It might explain the immediate effect of a topical cream on your skin.
While these applications are at least a decade away from becoming mainstream, improvements in wearable devices, sensors, digital health solutions and IoT are spurring the development and use of these digital biomarkers, said Isernhagen.
In fact, it is one of the areas of technology that SGInnovate is investing in, said Hsien-Hui Tong, Executive Director of Venture Investing, SGInnovate, during the webinar.
2. Autonomous vehicles will play a larger part in commercial and industrial operations
This year, the autonomous vehicles industry will focus more on commercial and industrial applications, and less on personal use. This shift is driven by the disruption of manufacturing processes and supply chains during COVID-19.
The pandemic has spurred innovation not only in last-mile logistics, but also in the middle mile. That is where autonomous vehicles come into play. Repetitive movements and unchanging routes, such as goods transport from the port to the warehouse, can be done by automated fleets.
More autonomous vehicles and machines will also emerge within warehouses this year. “In the warehouse, a large company can customise the robots just for what they need. The benefit is clear and obvious,” said Peter Norvig, Director of Research, Google, speaking at The Next Step for Deep Tech in Asia webinar.
The necessary technology is falling into place. Sensors — a crucial piece of the robotics puzzle — are becoming cheaper. Additionally, AI is becoming more precise, and 5G more widespread.
3. All AI systems suffer from weaknesses — and it is time to solve them
It is impossible to build accurate and effective AI and machine learning models without clean data. This problem is beginning to be solved with the development of more affordable and more precise sensors. With a greater volume of high-quality sensors, systems can collect more amounts of data at an improved accuracy rate.
However, as AI models analyse more data, they require more computational power. Quantum computing can provide such paper but will take many more years to become commercially widespread. The good news is we now see increased corporate interest in this area following Google’s claim of quantum supremacy.
With more computing power, organisations can deploy large multi-modal models, which combine different types of data — language, speech, image, video, and more. This greater variety of parameters will enable AI models to interpret data more accurately.
Amid all these developments, though, there is one problem the industry can no longer afford to ignore — AI’s massive consumption of energy. “At the rate AI is going, it’s going to be a very carbon-intensive way of running businesses,” said Tong.
In response, the Deep Tech sector is designing low-power AI processors that use less electricity. 2020 saw breakthroughs in the testing and commercialisation of these chips — a development that will gain momentum this year.
Lastly, data management and privacy can be a challenge, as we have seen in the controversies surrounding the way businesses collect and use data. Federated learning can help users keep their data, but such a system requires significant processing power on local devices. AI chip advancements can help make this a reality.
4. Organisations everywhere are embracing digitalisation, but the kind of tech needed varies widely across APAC countries
Asia’s developing markets are embracing technology because it opens access to new products and services. Consider, for example, how mobile wallets are serving customers who have never had bank accounts, providing them access to loans, insurance, and digital money transfers.
But while the pandemic has accelerated digital adoption across Asia Pacific, the progress of business digitalisation and automation differs widely across the region.
“If you're in Singapore, for instance, you are at a much later stage of digitalisation, and you can afford to look at investments in some of the more sophisticated AI techniques. In other parts of Asia, just the basic aspect of digitising your data inputs is a challenge,” said Tong.
Businesses in more developed markets have come to see automation as essential to business continuity plans, said Sinuhé Arroyo, Founder & CEO, Taiger, during The Next Step for Deep Tech in Asia webinar. Taiger is a global AI company headquartered in Singapore.
That means we can expect to see greater adoption this year of tech solutions that enable automation.
“We see clients investing in machine learning, AI, and robotic process automation solutions to automate [activities],” said Christopher Warren, Managing Director – Strategy & Consulting, Accenture, during The Promise of Deep Tech webinar.
He added that by adopting such technology, organisations can improve their performance and create more bandwidth to innovate even further.
The promise of a better year ahead
As a world reeling from a pandemic tries to get back on its feet, it can find hope in Deep Tech developments, especially innovations in healthcare and commerce, to help us create a more resilient, sustainable and secure future.
Interested to learn more breaking insights on the latest and greatest about Deep Tech? You can check out SGInnovate’s monthly event line-up here.
Share this with your network!