Article | January 28, 2021
Since the internet became popular, the way we purchase things has evolved from a simple process to a more complicated process. Unlike traditional shopping, it is not possible to experience the products first-hand when purchasing online. Not only this, but there are more options or variants in a single product than ever before, which makes it more challenging to decide.
To not make a bad investment, the consumer has to rely heavily on the customer reviews posted by people who are using the product. However, sorting through relevant reviews at multiple eCommerce platforms of different products and then comparing them to choose can work too much. To provide a solution to this problem, Amazon has come up with sentiment analysis using product review data. Amazon performs sentiment analysis on product review data with Artificial Intelligence technology to develop the best suitable products for the customer. This technology enables Amazon to create products that are most likely to be ideal for the customer.
A consumer wants to search for only relevant and useful reviews when deciding on a product. A rating system is an excellent way to determine the quality and efficiency of a product. However, it still cannot provide complete information about the product as ratings can be biased. Textual detailed reviews are necessary to improve the consumer experience and in helping them make informed choices. Consumer experience is a vital tool to understand the customer's behavior and increase sales.
Amazon has come up with a unique way to make things easier for their customers. They do not promote products that look similar to the other customer's search history. Instead, they recommend products that are similar to the product a user is searching for. This way, they guide the customer using the correlation between the products.
To understand this concept better, we must understand how Amazon's recommendation algorithm has upgraded with time.
The history of Amazon's recommendation algorithm
Before Amazon started a sentiment analysis of customer product reviews using machine learning, they used the same collaborative filtering to make recommendations. Collaborative filtering is the most used way to recommend products online. Earlier, people used user-based collaborative filtering, which was not suitable as there were many uncounted factors.
Researchers at Amazon came up with a better way to recommend products that depend on the correlation between products instead of similarities between customers. In user-based collaborative filtering, a customer would be shown recommendations based on people's purchase history with similar search history. In item-to-item collaborative filtering, people are shown recommendations of similar products to their recent purchase history. For example, if a person bought a mobile phone, he will be shown hints of that phone's accessories.
Amazon's Personalization team found that using purchase history at a product level can provide better recommendations. This way of filtering also offered a better computational advantage. User-based collaborative filtering requires analyzing several users that have similar shopping history. This process is time-consuming as there are several demographic factors to consider, such as location, gender, age, etc. Also, a customer's shopping history can change in a day. To keep the data relevant, you would have to update the index storing the shopping history daily.
However, item-to-item collaborative filtering is easy to maintain as only a tiny subset of the website's customers purchase a specific product. Computing a list of individuals who bought a particular item is much easier than analyzing all the site's customers for similar shopping history. However, there is a proper science between calculating the relatedness of a product. You cannot merely count the number of times a person bought two items together, as that would not make accurate recommendations.
Amazon research uses a relatedness metric to come up with recommendations. If a person purchased an item X, then the item Y will only be related to the person if purchasers of item X are more likely to buy item Y. If users who purchased the item X are more likely to purchase the item Y, then only it is considered to be an accurate recommendation.
In order to provide a good recommendation to a customer, you must show products that have a higher chance of being relevant. There are countless products on Amazon's marketplace, and the customer will not go through several of them to figure out the best one. Eventually, the customer will become frustrated with thousands of options and choose to try a different platform. So Amazon has to develop a unique and efficient way to recommend the products that work better than its competition.
User-based collaborative filtering was working fine until the competition increased. As the product listing has increased in the marketplace, you cannot merely rely on previous working algorithms. There are more filters and factors to consider than there were before. Item-to-item collaborative filtering is much more efficient as it automatically filters out products that are likely to be purchased. This limits the factors that require analysis to provide useful recommendations.
Amazon has grown into the biggest marketplace in the industry as customers trust and rely on its service. They frequently make changes to fit the recent trends and provide the best customer experience possible.
Article | February 10, 2020
We are a species invested in predicting the future as if our lives depended on it. Indeed, good predictions of where wolves might lurk were once a matter of survival. Even as civilization made us physically safer, prediction has remained a mainstay of culture, from the haruspices of ancient Rome inspecting animal entrails to business analysts dissecting a wealth of transactions to foretell future sales. With these caveats in mind, I predict that in 2020 (and the decade ahead) we will struggle if we unquestioningly adopt artificial intelligence (AI) in predictive analytics, founded on an unjustified overconfidence in the almost mythical power of AI's mathematical foundations. This is another form of the disease of technochauvinism I discussed in a previous article.
Article | March 9, 2021
For many, 2021 has brought hope that they can cautiously start to prepare for a world after Covid. That includes living with the possibility of future pandemics, and starting to reflect on what has been learned from such a brutal shared experience. One of the areas that has come into its own during Covid has been artificial intelligence (AI), a technology that helped bring the pandemic under control, and allow life to continue through lockdowns and other disruptions.
Plenty has been written about how AI has supported many aspects of life at work and home during Covid, from videoconferencing to online food ordering. But the role of AI in preventing Covid causing even more havoc is not necessarily as widely known. Perhaps even more importantly, little has been said about the role AI is likely to play in preparing for, responding to and even preventing future pandemics.
From what we saw in 2020, AI will help prevent global outbreaks of new diseases in three ways: prediction, diagnosis and treatment.
Predicting pandemics is all about tracking data that could be possible early signs that a new disease is spreading in a disturbing way. The kind of data we’re talking about includes public health information about symptoms presenting to hospitals and doctors around the world. There is already plenty of this captured in healthcare systems globally, and is consolidated into datasets such as the Johns Hopkins reports that many of us are familiar with from news briefings.
Firms like Bluedot and Metabiota are part of a growing number of organisations which use AI to track both publicly available and private data and make relevant predictions about public health threats. Both of these received attention in 2020 by reporting the appearance of Covid before it had been officially acknowledged. Boston Children’s Hospital is an example of a healthcare institution doing something similar with their Healthmap resource.
In addition to conventional healthcare data, AI is uniquely able to make use of informal data sources such as social media, news aggregators and discussion forums. This is because of AI techniques such as natural language processing and sentiment analysis. Firms such as Stratifyd use AI to do this in other business settings such as marketing, but also talk publicly about the use of their platform to predict and prevent pandemics. This is an example of so-called augmented intelligence, where AI is used to guide people to noteworthy data patterns, but stops short of deciding what it means, leaving that to human judgement.
Another important part of preventing a pandemic is keeping track of the transmission of disease through populations and geographies. A significant issue in 2020 was difficulty tracing people who had come into contact with infection. There was some success using mobile phones for this, and AI was critical in generating useful knowledge from mobile phone data.
The emphasis of Covid tracing apps in 2020 was keeping track of how the disease had already spread, but future developments are likely to be about predicting future spread patterns from such data. Prediction is a strength of AI, and the principles used to great effect in weather forecasting are similar to those used to model likely pandemic spread.
To prevent future pandemics, it won’t be enough to predict when a disease is spreading rapidly. To make the most of this knowledge, it’s necessary to diagnose and treat cases. One of the greatest early challenges with Covid was the lack of speedy, reliable tests.
For future pandemics, AI is likely to be used to create such tests more quickly than was the case in 2020. Creating a useful test involves modelling a disease’s response to different testing reagents, finding right balance between speed, convenience and accuracy. AI modelling simulates in a computer how individual cells respond to different stimuli, and could be used to perform virtual testing of many different types of test to accelerate how quickly the most promising ones reach laboratory and field trials.
In 2020 there were also several novel uses of AI to diagnose Covid, but there were few national and global mechanisms to deploy these at scale. One example was the use of AI imaging, diagnosing Covid by analysing chest x-rays for features specific to Covid. This would have been especially valuable in places that didn’t have access to lab testing equipment. Another example was using AI to analyse the sound of coughs to identify unique characteristics of a Covid cough.
AI research to systematically investigate innovative diagnosis techniques such as these should result in better planning for alternatives to laboratory testing. Faster and wider rollout of this kind of diagnosis would help control spread of a future disease during the critical period waiting for other tests to be developed or shared. This would be another contribution of AI to preventing a localised outbreak becoming a pandemic.
Historically, vaccination has proven to be an effective tool for dealing with pandemics, and was the long term solution to Covid for most countries. AI was used to accelerate development of Covid vaccines, helping cut the development time from years or decades to months. In principle, the use of AI was similar to that described above for developing diagnostic tests.
Different drug development teams used AI in different ways, but they all relied on mathematical modelling of how the Covid virus would respond to many forms of treatment at a microscopic level.
Much of the vaccine research and modelling focused on the “spike” proteins that allow Covid to attack human cells and enter the body. These are also found in other viruses, and were already the subject of research before the 2020 pandemic. That research allowed scientists to quickly develop AI models to represent the spikes, and simulate the effects of different possible treatments. This was crucial in trialling thousands of possible treatments in computer models, pinpointing the most likely successes for further investigation.
This kind of mathematical simulation using AI continued during drug development, and moved substantial amounts of work from the laboratory to the computer.
This modelling also allowed the impact of Covid mutations on vaccines to be assessed quickly. It is why scientists were reasonably confident of developing variants of vaccines for new Covid mutations in days and weeks rather than months.
As a result of the global effort to develop Covid vaccines, the body of data and knowledge about virus behaviour has grown substantially. This means it should be possible to understand new pathogens even more rapidly than Covid, potentially in hours or days rather than weeks.
AI has also helped create new ways of approaching vaccine development, for example the use of pre-prepared generic vaccines designed to treat viruses from the same family as Covid. Modifying one of these to the specific features of a new virus is much faster than starting from scratch, and AI may even have already simulated exactly such a variation.
AI has been involved in many parts of the fight against Covid, and we now have a much better idea than in 2020 of how to predict, diagnose and treat pandemics, especially similar viruses to Covid. So we can be cautiously optimistic that vaccine development for any future Covid-like viruses will be possible before it becomes a pandemic. Perhaps a trickier question is how well we will be able to respond if the next pandemic is from a virus that is nothing like Covid.
Was Rahman is an expert in the ethics of artificial intelligence, the CEO of AI Prescience and the author of AI and Machine Learning. See more at www.wasrahman.com
Article | March 15, 2021
Stephen Hawking, one of the finest minds to have ever lived, once famously said, “AI is likely to be either the best or the worst thing to happen to humanity.” This is of course true, with valid arguments both for and against the proliferation of AI.
As a practitioner, I have witnessed the AI revolution at close quarters as it unfolded at breathtaking pace over the last two decades. My personal view is that there is no clear black and white in this debate. The pros and cons are very contextual – who is developing it, for what application, in what timeframe, towards what end?
It always helps to understand both sides of the debate. So let’s try to take a closer look at what the naysayers say. The most common apprehensions can be clubbed into three main categories:
A. Large-scale Unemployment: This is the most widely acknowledged of all the risks of AI. Technology and machines replacing humans for doing certain types of work isn’t new. We all know about entire professions dwindling, and even disappearing, due to technology. Industrial Revolution too had led to large scale job losses, although many believe that these were eventually compensated for by means of creating new avenues, lowering prices, increasing wages etc.
However, a growing number of economists no longer subscribe to the belief that over a longer term, technology has positive ramifications on overall employment. In fact, multiple studies have predicted large scale job losses due to technological advancements. A 2016 UN report concluded that 75% of jobs in the developing world are expected to be replaced by machines!
Unemployment, particularly at a large scale, is a very perilous thing, often resulting in widespread civil unrest. AI’s potential impact in this area therefore calls for very careful political, sociological and economic thinking, to counter it effectively.
B. Singularity: The concept of Singularity is one of those things that one would have imagined seeing only in the pages of a futuristic Sci-Fi novel. However, in theory, today it is a real possibility. In a nutshell, Singularity refers to that point in human civilization when Artificial Intelligence reaches a tipping point beyond which it evolves into a superintelligence that surpasses human cognitive powers, thereby potentially posing a threat to human existence as we know it today.
While the idea around this explosion of machine intelligence is a very pertinent and widely discussed topic, unlike the case of technology driven unemployment, the concept remains primarily theoretical. There is as yet no consensus amongst experts on whether this tipping point can ever really be reached in reality.
C. Machine Consciousness: Unlike the previous two points, which can be regarded as risks associated with the evolution of AI, the aspect of machine consciousness perhaps is best described as an ethical conundrum. The idea deals with the possibility of implanting human-like consciousness into machines, taking them beyond the realm of ‘thinking’ to that of ‘feeling, emotions and beliefs’.
It’s a complex topic and requires delving into an amalgamation of philosophy, cognitive science and neuroscience. ‘Consciousness’ itself can be interpreted in multiple ways, bringing together a plethora of attributes like self-awareness, cause-effect in mental states, memory, experiences etc. To bring machines to a state of human-like consciousness would entail replicating all the activities that happen at a neural level in a human brain – by no means a meagre task.
If and when this were to be achieved, it would require a paradigm shift in the functioning of the world. Human society, as we know it, will need a major redefinition to incorporate machines with consciousness co-existing with humans. It sounds far-fetched today, but questions such as this need pondering right now, so as to be able to influence the direction in which we move when it comes to AI and machine consciousness, while things are still in the ‘design’ phase so to speak.
While all of the above are pertinent questions, I believe they don’t necessarily outweigh the advantages of AI. Of course, there is a need to address them systematically, control the path of AI development and minimize adverse impact. In my opinion, the greatest and most imminent risk is actually a fourth item, not often taken into consideration, when discussing the pitfalls of AI.
D. Oligarchy: Or to put it differently, the question of control. Due to the very nature of AI – it requires immense investments in technology and science – there are realistically only a handful of organizations (private or government) that can make the leap into taking AI into the mainstream, in a scalable manner, and across a vast array of applications. There is going to be very little room for small upstarts, however smart they might be, to compete at scale against these.
Given the massive aspects of our lives that will likely be steered by AI enabled machines, those who control that ‘intelligence’ will hold immense power over the rest of us. That all familiar phrase ‘with great power, comes great responsibility’ will take a whole new meaning – the organizations and/or individuals that are at the forefront of the generally available AI applications would likely have more power than the most despotic autocrats in history. This is a true and real hazard, aspects of which are already becoming areas of concern in the form of discussions around things like privacy.
In conclusion, AI, like all major transformative events in human history, is certain to have wide reaching ramifications. But with careful forethought these can be addressed. In the short to medium term, the advantages of AI in enhancing our lives, will likely outweigh these risks. Any major conception that touches human lives in a broad manner, if not handled properly, can pose immense danger. The best analogy I can think of is religion – when not channelled appropriately, it probably poses a greater threat than any technological advancement ever could.