Coronavirus

Are we at the AI inflection point 2.0?

The evolution of AI from a cool toy to an essential technology appears to be inevitable in the face of the Covid-19 pandemic

Updated: Jul 21, 2020 01:31:39 PM UTC

Srivatsan Laxman is Founder & CEO of True Lark, Aniruddha Banerjee is Co-Founder of SwitchOn, and Manish Singhal is Founding Partner at pi Ventures

SM_AI-Op-ed
Image: Shutterstock

In the past six months, the world as we know it has been upended. While the jury is out on how it the Covid-19 pandemic will end and what aspects of life will change forever, there are a few things we know for certain that are going to be different in the new world we are heading into:

Remote is the new local
The shift to remote will have a transformative impact on all aspects of life including travel, hospitality, work, buying behaviour, manufacturing and the shared economy. All forms of automation will see massive tail winds.

Digital everywhere
The digital economy will create exponentially more data. As the world goes remote, more people will collaborate and transact online, leading to an exponential increase in the amount of data generated. This will give rise to unprecedented new data insights and data-driven applications

The evolving AI landscape SM_AI-Op-ed

Communities around the world are establishing new norms for social interaction and this is opening up exciting problem spaces for artificial intelligence (AI) and man-machine interfaces, across a wide range of industries. As the pandemic runs its course, businesses are adapting to changes in customer behaviour. They are also re-imagining their own staffing, collaboration and productivity models mandated by new cost structures and modified working conditions.

Industries in slow down
Some industries have seen an obvious and dramatic slowdown that may not see an easy turnaround in the near term; e.g., people are unlikely to be as willing as before to share space, vehicles and tools, whether for work or for leisure. They are also just as unlikely to rush back to attending sports, concerts or other such large events. Even the simple joy of a movie or a play, or of meeting friends at a bar, feels like a thing of the past now. These industries will be back at the drawing board, in search of product innovations to deliver new customer experiences.

Emerging use cases
Covid-19 will catalyse the emergence of AI automation in many new applications. The health industry will start to enable telemedicine, remote diagnostics and remote patient care technologies. AI-powered solutions will become a key tool to assess and track mental health. Pharma will use AI to narrow an otherwise prohibitively large search space for new drugs and vaccines, dramatically accelerating their time to market. Robots for cleaning, shelf-stocking and last-mile delivery will go mainstream. Drones will evolve from their military origins into tools for businesses and governments; for example, to deliver essential medicines, collect patient samples, for area surveillance and public broadcast. Smart collaboration tools will make remote working more effective. HR will use automation to pre-screen candidates, match resumes and help with cognitive and emotional assessments. Chatbots can walk employees through onboarding, and AI will can assist with monitoring employee health and safety.

Accelerating use-cases
Areas where AI has already established a foothold are seeing rapid and accelerated adoption even during the shutdown. Contact centres have turned-on AI to drive automation and efficiency in customer service. Voice interfaces for support, sales and marketing are becoming commonplace. Smarting from the collapse of traditional models and processes during the pandemic, supply chains are putting in place next generation AI models for planning and forecasting.

Digital commerce is adopting AI more widely (even by the smaller players) to improve product recommendations and personalisation, enable virtual assistants, detect fraud and fake reviews, content generation etc. The Covid-19 shutdown pushed many legacy business processes beyond peak capacity, exposing an over-reliance on manual procedures, such as mass rescheduling in the airline industry, unprecedented loan applications in banks, etc. These industries are now turning to hyper-automation that combines robotic process automation with modern machine learning to ensure they can better handle surges in the future.

Advances in core AI technologies
In a bid to service these and other emerging and accelerating use-cases, many open problems in AI will see heightened research activity. For example, we will see the maturing of emotion AI technologies to interpret and respond to human emotions based on non-verbal cues such as facial expressions, gestures, body language and tone of voice. Learning from small data sets using approaches like transfer learning, few-shot learning and multi-task learning, will become more accessible to application developers leading to broader adoption. These developments will be critical as most applications still suffer from very limited labelled training data. Large-scale content generation for games will be powered by advancements in reinforcement learning and other self-learning AI technologies.

Explainable AI, data security and privacy
As the world moves to embrace these far-reaching applications of AI, the importance of fairness, accountability and transparency of AI cannot be overstated. Today AI is beginning to drive more and more applications that define our health and safety. Understanding the mechanics of how AI operates (explainable AI) has become critical in domains, such as healthcare and legal. There needs to also emerge a more accessible understanding of privacy-preserving AI and federated learning algorithms, as these can very effectively address concerns around security and data privacy. Most importantly, we must as a society develop a broader consensus for the oversight of bias and ethics in AI. Without this, the AI juggernaut will most certainly fail to achieve its full potential.

Are we at inflection point 2.0 for AI?
With the advent of deep learning, many core building blocks of AI—image and speech recognition, object detection and tracking, language understanding etc—achieved error rates that enabled powerful new applications to be built for the real world. This inspired us at pi Ventures to build an AI focussed fund back in 2016. Our thesis was simple (see figure) - startups that use AI would lead to 10x differentiated products and become category leaders in due course.

Today, we really view this as a first inflection point for AI in the modern era. In the last couple of years, AI has unequivocally assumed the mantle of innovation in the world of technology. The disruptions to society in 2020 will only broaden its appeal and accelerate its adoption. Almost certainly, the Covid-19 pandemic will be the second inflection point for AI. Even as technologists focus on the advances in AI, governments around the world will prioritise the relevant regulatory oversight needed. The evolution of AI from a cool toy to an irreversibly essential technology appears to be inevitable.

SM_AI-Op-ed

About the authors:
Srivatsan Laxman is Founder & CEO of True Lark, Aniruddha Banerjee is Co-Founder of SwitchOn, and Manish Singhal is Founding Partner at pi Ventures

Post Your Comment
Required
Required, will not be published
All comments are moderated
Prev
Technocracy: The next upskilling opportunity for Bharat's bureaucracy
Next
Is India's economic recovery for real?