AI takes over, but while doing so it is assigned tasks with real world consequences before AI works properly. Several videos are alogorithmically produced, with names that are collections of tags. Such videos targeted to young population makes the system complicit in abuse. The fake news would not have been possible without Google’s programmatic advertising technology and Facebook’s propensity to tolerate fake accounts. The tag-filled names of the videos are designed to exploit YouTube’s search algorithms. The catchy headlines of the fake stories continue fooling Facebook’s clickbait detection algorithms. MIT scientists have developed Moral Machine to automate the ethical decisions a human driver makes on the fly — whether to hit a wall and kill the car’s passengers, including a young girl or run over an athlete and his dog crossing the street on a red light. The researchers used a website to ask people about moral choices. Then the data is aggregated. AI-based algorithm figures out a decision corresponding to the crowdsourced wisdom. The self-driving cars can make credible decisions on ethical dilemmas by implementing an alogorithm on the Moral Machine data set. But it is not the end-all solution.
Alogorithms may have improved but it will be a long time before they perform tasks that require human judgment, without some human figuring out how to game them.
Month: November 2017
Machine Learning Decisions
Machine learning works by setting up programmes that could consider multiple variables and then taking as input huge quantity of data. The programme sifts the data. It formulates its own rules and finds patterns and co-relations. It is in effect a blackbox.
When there is a loan or credit card application, the yes/no decision is often made by a machine. The cedit limit, interest rates, tenure and other details are also likely to be automated.
Nobody except the machine knows the logic behind a decision. It is not known whether the machine is transparent and fair.
A person’s savings is allocated by intelligent agents. No fly lists of passengers are made by running algorithims. AI is used to read body language, and to figure out sexual orientation.
The Europeon Union’s proposed General Data Protection Regulation (GDPR) will be enforced from 2018, and some sections of the law give citizens a right to demand explanations about decisions made by algorithms. Article 22 implies that a human element would have to be present in the chain if algorithms are run to make key decisions that affect individuals.
Marketing Research Reinvention
Traditional marketing research originated in advertising agencies in the 1950s, as an adjunct to the account planning function.Gradually, these MR departments blossomed as full-fledged MR companies, eg. IMRB of India’s genesis was in the HTA. Manufacturers set up MR departments to interpret data later. Vikram Sarabhai set up ORG to do retail audit of pharmaceuticals.
Of late, Gigna US has shut up their MR department and it has been replaced by Marketing Analytics department. The digital era is influencing MR not only abroad, but in India too. Instead of long consumer tracking surveys, we could have microsurveys and intercepts. Data scientists could be assigned the insight management. Instead of rear-view analytics of trditional MR, there would be focus on predictive analytics.
There are multiple data streams these days. There is traffic to the brand’s web site. There are social media. The buzz here has to be monitored. The comments in the blogs must be factored in. Traditionally, data from the syndicated research flowed in. Research was commissioned. Consumer behaviour was mapped. The focus was on past data. Marketing analytics calls for new set of skills. There has to be training datasets to help project intended behaviour. Experiments are run in real life. They are run in behavioral labs. Models are built and validated. All this is done without diluting the core MR and consumer behaviour.
MR is transforming into Marketing Analytics. Traditional skills and new age skills coverage.
Online Video
There is a space for high quality content on the mobile. Every major player is investing heavily in acquiring or commissioning content. The first round of growth was subscription driven.
The kind of shows that are being commissioned needs consideration. There is an increase in creating more complex storytelling. The shows for online are made differently from those on the TV. TV is driven by the consumer research and ratings. The data that powers online content decision is different. It depends on deep consumer insight and expectation. Though successful TV shows can be adapted online, there are limitations. Producers of movies upload close to 30-40 promotional videos on YouTube, as against 2-3 previously, as there is co-relation between engagement with the videos and the box office success of a movie.
Online offers more freedom creatively and commercially. Online can explore a variety of genres such as crime, mature relationships which do not fit the daily soap format on TV. In the US, the subscriptions or pay TV drive the content. In India it is ad driven and mass driven, as subscriptions or pay TV has not taken off here.
Some audience in India lapses from TV. They do catch up TV on line or view IPL like big events.
TV has sheer stickiness. It reaches a mammoth 875 million in India. It has continuous engagement. This is not there for an online series that ends in five or ten episodes. Ad rates in video are a fraction of those on the TV. Thus monetisation is the biggest challenge.