Artificial intelligence will not only take away jobs but also create them

One of Molière’s plays, Le Bourgeois Gentilhomme (The Middle-class Gentleman), features a social climber who is delighted to discover that he’s been speaking prose all his life. Most of us would feel that way about artificial intelligence (AI) if we only realised how often we use it.

The commonest connection is via smartphones. Smartphones have virtual personal assistants (VPA) like Siri, Cortana or Google Assistant. When you ask the VPA to do something, it uses AI to recognise your voice, decipher your accent, follow orders and answer questions.

One of AI’s breakthroughs has been the creation of VPAs, which need skills a child acquires easily but a computer finds difficult. Recognising voices and understanding spoken language with its hesitations and grammatical errors is tough. Handwriting is even harder to interpret, and face-recognition is off-the-scale in complexity. But no VPA would work without this basic skill set. Other AIs do a heap of more challenging tasks. Some of these are beyond the competence of humans. Some of these are making many other professions redundant.

New technology always has this sort of disruptive effect. AI is bound to open up new opportunities, just like the automobile, the telephone and the internet did. But at this moment, the jobs under threat are more visible than the potential new opportunities. AI has already enabled an entirely new industry in fintech, allowing mobile wallet providers and payment banks to grab a slice of consumption.

Advances in processing power, the proliferation of Internet of Things (the vast network of devices, including vehicles and homes, connected to the internet) and new algorithmic techniques have combined to give AI and Machine Learning (AI and ML are often used as synonyms) a boost. 

We generate more data than ever. Greater processing power helps crunch that data and new ML techniques help computers optimise and analyse it, and learn better.

Although AI mimics human intelligence, computers “learn” differently. One great training ground is games. The ancient Eastern game of Go is much more complicated than chess. Chess has 20 possible opening moves; Go has 361. In 2016, an AI called AlphaGo started beating world champions. Its successor, AlphaGo Zero, is stronger.

In most AI, humans set basic guidelines and leave the AI to sift through vast quantities of data to improve understanding. For example, autonomous cars are shown images and told, “This is a traffic light, this is a billboard, this is the back of a truck”, and fed millions of related images. Handwriting- and image-recognition programs work similarly, with humans identifying samples to help the machine learn about variations in handwriting and in faces.

The first AlphaGo had humans teaching strategies and setting it to analyse millions of sample games played by humans. “Zero” was just taught the rules and told to play against itself. In a few days, it had discovered everything humans knew, and developed new insights as well. This auto-didacticism was quicker and less data-dependent. The technique could be adapted to deal with problems humans can define but not easily solve, such as gene-sequencing and protein-folding.

AI makes many tasks easier and it opens up new possibilities. New entrepreneurial opportunities will arise as humans figure this out. That will involve an inverse process: instead of programming AIs to do things, human beings will have to learn from AIs and work with AI.

It isn’t only STEM-based businesses — those built on science, technology, engineering and maths — that will thrive. AI can compose brilliant music in multiple styles; it can script complicated games, do great graphics and design superb buildings. Smart people who figure out how to use such capabilities could be big gainers.

However, there are huge challenges in reskilling. IBM has conducted studies such as “Upskilling India”, which surveyed academics, corporate recruiters, venture capitalists and educationists. The study highlighted that 60 per cent of global executives expect that employees will need new, different skills to be successful. About 70 per cent of India’s venture capitalists indicated that start-ups cannot find employees with the right skills. The sheer scale of how things will change implies that the entire education system will need to be rejigged.

Let’s look at the sectors where AI has been transformational. (In fact, it’s hard to think of areas where AI has not been a game-changer.) Military applications abound, of course. Drones and robots and locational services are now moving onto civvy street. AI is also being deployed in security and law-enforcement where face-recognition and data-sifting ability makes it very useful.

A range of low-level white collar tasks, and some high-level ones, have been taken over by intelligent agents and chatbots. Apart from secretarial and clerical services, AI is also helping with personal finance advice and basic portfolio management. It enables tech troubleshooting and support. This has led to the imminent demise of conventional call centre operations.

AIs today trade financial assets, too — most trades are done by autonomous programs. Where arbitrage opportunities arise in micro-seconds, human reflexes are outmatched. Nor can any human sift through mounds of financial data in real time to identify trading patterns.

For technology services companies such as Persistent Systems, it has opened a whole new world. “We [now] use text and image processing on photos of damaged cars at accident sites to determine the estimated cost of repair and improve insurance claims processing and policy underwriting,” says Abhay Pendse, chief architect, Persistent Systems. 

The IT industry itself has seen a vast churn as AI has ushered out the era of manpower-intensive coding and tech-support. Further ahead, new areas of competence will include hands-on programming expertise to fully utilise AI/ML capabilities. Data analysts, data scientists and data engineers must understand customer problems to find solutions. At the top level, technology experts will have to provide long-term vision.

The savings and productivity gains may be terrific. Prashant Pradhan, chief technology officer and chief developer advocate, IBM India/South Asia, says that AI could cut operational costs in maintaining telecom networks and trouble-shooting by one-third. In addition, productivity gains and even the generation of new revenue streams may be possible if telecom service providers can use data to find new ways to leverage existing networks. 

In health care and medicine, AI today enables early diagnosis and treatment of unusual cancers. It crunches through DNA analysis and gene-sequencing and figures out molecular combinations to discover new drugs. 

When it comes to oncology, IBM’s Watson is at the cutting-edge of diagnosis and treatment, working with many medical institutes, including Manipal Hospitals. Watson has a huge database of cancer scans, papers, records and case studies which it is constantly upgrading. It can speak to specialists in a natural or ordinary language, which means that it can help diagnose from stated symptoms. It can compare MRI scans and reports with its database. It can crunch genomic data to suggest treatments tailored for specific individuals. (However, Watson must still rely on a consulting panel of specialists to flag cases.)

Understanding patient intent is another area where AI is proving useful, says Rizwan Koita, CEO of CitiusTech, a health care-focussed IT services provider. “AI-driven chatbots are able to interpret ‘patient intent’ by analysing language and conversation patterns,” Koita explains. For example, they can identify emergency-care situations by correlating the use of specific words such as “ambulance”, “urgent” or “hurry”. 

As in health care, AI is also making inroads into the legal profession. It is already taking over legal research, eliminating the need for legal researchers who spend hours poring over case files. But when it comes to presenting a case and interpreting the law, lawyers and judges will still be needed.

There appears to be no limit to AI’s reach. It is no secret that Facebook and Google use image-recognition programs to classify pictures by provenance, check for copyright, inappropriate content and privacy violation. AI programs can go further and analyse video for individual characteristics like gait and body language, and match those to known persons. Programs such as “Silent Talker” work like lie detectors by looking at microscopic changes in facial expression when people answer questions. 

There are also some disturbing possibilities that AI throws up — invading or compromising a person’s privacy being the most obvious one. An AI program from Stanford seems to guess with remarkable accuracy a person’s sexual orientation by just looking at the profile picture. Given that LGBT orientation is a criminal offence in many jurisdictions, including in India, this could be used to harass or penalise people.

The more esoteric uses of AI involve sexbots that respond humanly to a conversation. The AI politician, Sam, is another oddball innovation. Designed by New Zealander Nick Gerritsen, Sam is driven by “the desire to close the gap between what voters want and what politicians promise, and what they actually achieve.” He’s a good fact-checker who is semi-serious about standing for elections.

So how do human beings retrain to live in this new world where so many jobs are under threat?  

In health care, CitiusTech’s Koita says, “Physicians, nurses, administrative staff, quality teams and risk-management professionals will need to build skills around using technologies like big data, Cloud and cognitive technologies.”

Data scientist Gauri Shah of Persistent Systems adds, “It is a fact that AI/ML will replace some human jobs and supplement many, but it will also continue to create new job opportunities over time.”

There is no doubt that people will find new ways to use AI. One way of looking at it is this: AI has given us a lot more in the way of free time by automating monotonous, time-consuming tasks. As a species, we will now have to figure out what we want to do with it.

GET SMART

Photo: iStock

Those who have used typewriters will recall how tedious spellcheck and corrections by hand used to be. Word processor programs made the entire process much easier by automating it. That reduced the need for skilled stenographer-typists. But it enabled easier communications and created the phenomenon of blogging. While stenos who did boring, repetitive tasks lost their jobs, people with a talent for expressing themselves received new tools and platforms. 

AI is doing similar things — on a much larger scale and at higher levels of abstraction — to a huge range of professions. Paradoxically, information technology has been among the hardest hit. Low-level coding needs are much reduced and Cloud-based programs and databases need less oversight. But smart thinkers who appreciate the possibilities have created many new uses for AI. 

The transport industry is at the cusp of transformation. Driverless trucks and cars are already on roads in the US and Europe, and millions of professional drivers could become redundant in a few years. On the other hand, cars and trucks could be transformed into mobile offices or even bedrooms with completely different interior designs. The vehicle itself could be deployed 24x7. That could totally alter commuting experiences as well as car ownership versus rental patterns. There will be big opportunities here for smart entrepreneurs.

In other areas, AI combined with virtual reality can create games like Pokémon Go. Musicians can generate different arrangements for their compositions with different voices and instrument. Architects can visualise designs in fantastic detail. Sports coaches can use the same technologies to figure out better ways to hit balls. Physicists can simulate cosmological theories. There’s no end to possible new applications – the only limiting factor would be our imagination.