Cognitive computing-a force multiplier

“Is it possible for a computer to become human like?”  We have seen a computer HAL 2000 speak fluent English, experience jealousy and do away with the spaceship crew to prevent its own termination in Stanley Kubrick’s movie 2001: A Space Odyssey. In Her, a 2013 science fiction film written, directed, and produced by Spike Jonze, set in the near future, the film’s protagonist Theodore Twombly purchases an operating system with Samantha, a virtual assistant who evolves from being a competent assistant to a constant companion.

They bond over their discussions about love and life and develop a relationship that turns intimate. We do not have Samantha’s at the moment, but we have pre-programmed virtual assistants in the form of Siri and Cortana. Work in cognitive computing is progressing and a computer program Emily Howell released its first album in 2010, and another program wrote its first movie, Sun Spring. IBM’s Watson recently forayed into Hollywood by creating movie trailer of the science film Morgan in a record time. Traditionally, making a movie trailer is labour intensive and takes between 10 to 30 days to complete but IBM’s Watson took about 24 hours to sort through footage to come up with an exciting trailer.

The ability of humans to think is amazing. The creation of a computer system that can think and reason like humans do is the aim of cognisable computing. The end goal of cognisable computing is simply to simulate thought processes in a computerised model. Therefore, cognitive computing is a combination of computer and cognitive science. It may require several AI technologies for a computer system to build cognitive abilities in a computer system such as machine learning, deep learning, neural networks and sentiment analysis.

Old method of crime investigation involved collection of evidence from the field and forensic reports, the new method that is emerging is data driven, combining machine learning algorithms with big-data. The more data we feed into a computing system, the more a system will digest and assimilate leading to better insights. This method is being embraced by law enforcement agencies all over the world including the US. Cognitive systems, unlike humans, don’t sleep and have no emotions. They have capacity to read millions of documents, make connections and reveal patterns that law enforcement officials will find hard to catch sight of. When police arrive at a scene, they may not have adequate information to deal with the situation at hand, but cognitive systems can in a matter of minutes piece together disparate information and provide crucial information which could warn them of the presence| of explosives or past violence. Therefore, cognitive computing in law enforcement could compensate for human shortcomings by assisting them with a better decision making.

IBM’s cognitive supercomputer Watson, whose enormous data processing prowess helped it win “Jeopardy”, has been analysing data from health care to engineering. It has now gotten started on policing. The police firing during the Sterlite agitations in Tuticorin sparked a national discussion on police overreaction and blatant use of force by law-enforcement. There was rancour and rhetoric drawing many people into a debate over the police excesses but there was none who had the data to give the correct insight regarding the actual truth. They threw a lot of theories out in the media speculating, who deserves the blame. But we can’t rely on anecdotal information as there is a lot of emotion behind it. To get past the emotion and find the truth, we would need to get hold of a mound of data that the incident would have generated.

Today, Watson and cognitive computing systems like it, which are emerging on the law-enforcement space, can process all the data emanating from such incidents and give real insights faster. If we get Watson on the case, it could get somewhere close to the bottom of the truth, no matter how deep it’s buried - in considerably less time. IBM’s CopLink software is already helping police with data analysis. For instance, investigation of abduction of 6-year-old girl in Tucson, Arizona, out of her home generated 15,000 pages of reports, statements, lab reports etc., but the software could trudge through all the information and data, and generate leads, as opposed to speculation or analysis based on conjecture and supposed truth.

A Hindu religious festival is now on in the temple of Varadaraja Perumal, Kanchipuram, Tamil Nadu since July 1, 2019. The deity, made of fig tree wood, rests in the temple tank and which the temple priests draw out once in 40 years to enable the devotees to offer prayers to it for 48 days. Hence, the temple is popular by the name Athi Varadar temple. The last time priests placed the deity for devotees to offer obeisances was on July 2, 1979 and earlier on July 12, 1939. In 1979, the priests threw only one gate of the temple open, and the pundits told me that the number of devotees during the first 44 days averaged less than 10,000 per day. This time around the assessment before the start of the festival was not over 30,000 per day. Police as a measure of abundant caution made arrangements to control a crowd of one lakh devotees per day. But thanks to the TV Channels, social media, internet, the word of mouth, improved road and transport facilities, easy affordability of cars, increased materialism driving people towards spiritual solace and several such factors caused an unprecedented number of devotees to throng to the temple throwing the entire town out of gear. Ever since the start of the Athi Varadar festival till date, over 36 lakh devotees have paid obeisance to Lord Athi-Varadhar. On July 14 and 17 July alone, the temple witnessed 2.5 lakh and 2.75 lakh devotees respectively. On Sunday (July 27), the temple again witnessed a stream of over two lakh devotees. Handling such huge crowds for 48 days is a massive challenge for any administration. Under such a challenging situation, the administration is doing a great job despite enormous hardships. A cognitive crowd prediction system such as Watson by crunching available data such as the auspiciousness of the day, weather, the day i.e. working day or holiday, inflow of vehicles into the city, railway bookings from other states into Kanchipuram could have provided valuable information to the law enforcement for making deployment decisions.

Cognitive computing enables a digital policing environment that can help the police and the citizens immensely. Citizens will be able to update the police from anywhere anytime from any device using many channels such as photos, videos, voice calls. Word processing documents, emails, videos, images,audio files, presentations, webpages, social media and many other data formats often need manual tagging with metadata before they feed to a computer for analysis and insight generation. The principal benefit of using cognitive analytics over traditional big data analytics is that such datasets need not be pre-tagged. All the data coming in could help develop a more seamless way to solve crime and increase safety.

Besides providing specific insights to officers in the field, cognitive capabilities can help the investigators immensely in solving criminal cases. For example, the Law Enforcement Analysis Portal (LEAP), composed of a confederation of U.S. law enforcement agencies, is propelling a cognitive-enabled service that helps officers identify locations where we can find suspects. In the days gone by, when officers investigated cases, they had to count on information and evidence unearthed through painstaking investigation. Using cognitive computing abilities, LEAP can dig up data from suspects I-T records, property registration details, passport application information, speeding tickets, credit card, social-media interactions, internet browsing history, and licence plate databases. By having access to more data, police could piece together a more accurate picture of the most likely place they might locate the suspect. For example, an online passport application form details the suspect’s address, his marital status, his residential address, his mobile number, etc.

In the UK, using previous data, the Cambridge Police Department’s crime analysis unit, taught the Series Finder to detect patterns in the burglaries for over a decade in the university town of Cambridge. The factors which they incorporated to determine a Modus operandi comprised mode of entry (front door, back door
window, through the roof etc), time of the day, the day of the week, month, type of property, area and
proximity to other burglaries.  Using criminal patterns, the Series Finder could detect most of the reported cases and help in recovery as well. In addition, it also excluded some crimes that they ought to exclude, leading to a better idea about the actual suspect. This is something close to the pre-cogs which forecast crimes even before they happen in a famous story titled “Minority Report “by K.D. Phillips which Hollywood made into a movie of the same title. The machines today are not just reading but also interpreting information.

Although cognitive computing systems in the long run could replace humans in some situations, the systems would still need humans to direct them. Computing systems will only be able to augment human capabilities but not replace them. Just because supercomputers can read medical reports doesn’t mean that we won’t need doctors. The personal assistants we have on phones such as Siri are not true cognitive systems they respond only to pre-programmed set of responses but we are arriving at a time when our phones, cars, our computers will give us a real thoughtful response than a pre-programmed one. Computers will think like us and make accurate predictions, draw conclusions, and augment human capabilities in new ways.

We have developed machines that can operate autonomously, such as self-driving cars, robots, etc. But these autonomous machines or cars need to be told by human consciousness to drive from point A to point B or do whatever they can do. We could consider machines cognitive only when they decide by themselves and do what they wish to do rather than what we have programmed them to do. We are being told that Singularity is inevitable, does that portend cognitive machines?

(Dr Jayanth K Murali, IPS, is ADGP (Law and Order) Tamil Nadu. He can be contacted at www.jayanthmurali.com )