Thursday 21 September 2017

Five bold predictions on Artificial Intelligence

NEW YORK, August 21, 2017

Artificial Intelligence (AI) will have an enormous influence on the way we live our lives in the next decade. From enabling hyper-personalisation to saving huge amounts of time on routine tasks, these new tools will fundamentally shift the way we interact with technology in our day-to-day lives, says a report.

In the report titled "The Future of Artificial Intelligence in Consumer Experience", AT&T Foundry makes five bold projections that showcase how AI will impact the consumer experience in coming years.  

The market for AI has grown tremendously in the last couple of years. Right now, approximately 1,500 companies in North America are developing AI applications including leading companies such as Microsoft, IBM, Google, and Amazon. This number is only expected to grow as institutions embrace AI’s ability to increase productivity through intelligent automation, labour and capital augmentation, and innovation diffusion through AI partnerships, the report said.

Experts also forecast that annual global revenue from AI products and services will grow from $643.7 million in 2016 to as high as $36.5 to $100 billion by 2025.

What is AI?
Artificial intelligence is the development and use of computers to perform tasks that traditionally require human intelligence, such as visual perception, speech recognition, and language translation. With AI, computers learn from data sets to understand underlying data structures and uncover procedures to make the correct use of the data.

A crucial differentiation is between AGI (artificial general intelligence) and machine learning. AGI refers to machines generally being able to carry out tasks typically performed by humans. Machine learning refers to all techniques that allow computers to “learn without being explicitly” programmed. A particular application of machine learning is deep learning, which allows for a computer to recognise patterns from both labeled and unlabeled datasets.

However, AI is broader than machine learning or its subset, deep learning. Other AI techniques include search, symbolic reasoning, logical reasoning and statistical techniques that aren’t explicitly deep learning based. A truly comprehensive AI makes use of all techniques available, though the enterprise landscape is heavily focused on machine learning at the moment, the report said.


1. Humans have more room to be human
AI will fundamentally alter what it means to be human in this world. As algorithms automate away routine decision-making, people will see both an increase in the amount of time they have for tasks around critical thinking and creativity, as well as an increase in what they can do with that time.

With autonomous vehicles, it’s easy to imagine a future where consumers do not spend time physically driving a car or scheduling a pick up. These actions will be automated based on behavioural patterns and work routines, leaving time and space for “higher order thinking.” Car manufacturers and ride sharing companies are challenged to reinvent the space within the vehicle based on how users will want to spend their time, when attention at the wheel is no longer required. Will users want to be entertained, do work, sleep, or socialise?

Businesses within the autonomous vehicle space will need to rethink the transportation experience according to how users want to spend this time.

Higher order capabilities will be augmented by AI to extend natural barriers for conceptualisation and execution of creative endeavours. Even today, visual arts and music creation tools use AI to make the composition process more intuitive to the creators’ intentions and goals. AI will be a collaborator for humans as they are inspired to push past current boundaries and test new concepts across all areas of innovation.

As described by Jurgen Schmidhuber, co-founder of deep learning startup NNAIsense: “Our formal theory of fun even allows us to implement artificial curiosity and creativity, to build artificial scientists and artists.”

Ensuring that the users have the technical literacy to engage with -- and develop alongside -- AI tools will become a focal point for both traditional and career education. Meanwhile, products that go beyond functional solutions and allow users to quickly do the most interesting, exciting and meaningful things with their time, will win the consumer market, the report said.

2. Be everywhere as data is everywhere
AI applications are only as strong as the depth and quality of the data behind them. Currently, many organisations rely on proprietary data sets to train their applications, while interoperability and data-centric applications are secondary concerns. This limits the knowledge and processes a system can develop, especially with regard to customer behaviour and preferences. Increasingly, institutions are recognising that their development will be bolstered when they make their datasets available and usable by other organisations.

Shared data will enable platforms to communicate easily with one another, allowing user preferences to translate across devices and applications. The best consumer solutions will be those that effectively make use of all available customer data in a safe and secure manner to create comprehensive customer profiles and to accurately predict their needs.

There are a growing number of personalised products in the market that are the result of individual consumers’ aggregate purchasing and browsing data. As predictive analytic technology becomes more complex, companies will be able to create ideal products for each consumer before consumers can even request them. Customers will become accustomed to, and demand, applications that understand their preferences seamlessly.

Organisations that do not embrace data openness will be left behind by those who can act on quality consumer insights from sources both within and outside of their walls. This process includes not only exposure of current data, but the development of applications that are focused on secure transfers of centralized data from the beginning. Companies that set their infrastructure up for secure interoperability will be able to meet consumer demand for intuitive interfaces and applications more quickly.

For the consumer, this means less time spent on programming preferences and a more fluid user experience across devices. It also entails a public discussion around privacy rights and what types of data are appropriate for enterprise use, as a singular company’s user information will become everyone’s information.

3. Connectivity instantly powers your own adventure
Beyond YouTube and Snapchat, consumers will use newer technologies like augmented and virtual reality, haptic feedback for sensory experiences, and algorithmic storytelling. As consumer demand for these immersive interactions increase, underlying network and infrastructure will require more strength.

AI has three major impacts on connectivity networks: 1) it allows for accurate traffic and pattern analysis to troubleshoot problems as they occur, in turn allowing for 2) a constant state of connectivity that’s optimised for any experience across any set of devices, and 3) pulls disparate information from multiple channels to simplify and quickly contextualise what users need.

Messaging as a Platform, or MaaP, has emerged as a leading platform for AI innovation. Faced with a plethora of mobile applications to choose from, customers are increasingly using messaging as a simple channel through which to efficiently access information, entertainment, and amplify every day interactions. Leading global trade body GSMA has an active project that fosters an ecosystem of chatbots to support conversational commerce and intelligent assistants through messaging, in which AT&T is a contributor.

Intelligent networks will be able to predict network optimisation demands and grid failures before they occur, and proactively implement solutions so that the consumer never experiences the problem. The resulting state of robust connectivity will allow brands to make use of real-time user data to produce more relevant experiences both on devices and between devices.

Imagine a world where media platforms instantaneously update contextual user preferences across devices without having to ask for user input at any point. Voice-enabled devices pick up cues quickly and execute seamlessly on the optimised connection network. Televisions predict what you want to watch before you even turn them on. With the help of AI, the end user experience will be a smooth interaction across applications and devices.

4. Consumers go from one click to zero clicks
Experts predict that in five years, 85 per cent of business relationships with consumers will be managed without human interaction. Though there’s already an
industry-agnostic mantra on making business decisions with consumer needs as the primary driver, today’s understanding of the consumer is nowhere near what it will be when AI becomes mainstream. Due to ongoing data analysis at both individual and aggregate consumer levels, brands will create experiences that naturally integrate with each consumer’s day-to-day lives.

Consumers will no longer need to change their daily schedules or patterns of communication in order to get what they need from their favorite brands. Due to a deep comprehension of the customer, brands will provide sublime experiences catered to users’ behavioral patterns. Everything from shopping to driving will draw from user behaviour to become highly pertinent and personalised to the end consumer. Intelligent prediction and optimisation will allow the consumer to feel that each branded product or experience is made just for them.

Corporations will be able to assess shopper inventories and consumer behaviours to predict what items will be needed and deliver them directly to consumer homes before they even realise they’re running low. Even asking for help on an order will become more natural as platforms infuse AI with emotions to empathise with consumer needs and communicate solutions conversationally across interfaces.

With self-driving cars, consumers’ preferred routes and in-vehicle entertainment choices will draw from past behaviours (even those on other devices) to optimise both daily commutes and cross-country road trips.

Brands will act as “personal concierges” for a consumer’s needs, knowing what they want and how and when they want it before the consumer has to say anything at all.

5. Ethical AI controls for bias
The ethical impact of AI and automation is far reaching. One aspect corporations and institutional developers have direct control over is bias mitigation within their datasets and applied algorithms. In order to make AI work for all of society, application and platform developers will have to become conscious about the ways unmonitored data sets can aggregate and apply traditional human biases such as racism and sexism, and work towards mitigating these effects in the long term.

In nascent AI applications, we have already seen ethical dilemmas with computer vision technologies not accurately recognising the physiological characteristics of certain races and amplifying systemic biases contained within existing data sets. For example key risk assessment algorithm used by the US criminal justice system was found to be biased against black people in 2016. When used without human oversight, this algorithm would imprison black people at greater rates and for longer sentences than it would for the same crimes committed by other races.

As users come from all backgrounds, the successful integration of AI into society is dependent on producers’ ability to account for algorithmic biases before they become mainstream. Corporations will need to be cognizant of the potential for bias at every step of developing AI solutions, from capturing data optimized for all types of consumers to implementing parameters that account for the diversity of users’ socio-economic backgrounds, the report said.

As Tim Chang of the Mayfield Fund notes: “There are probably going to be data ethics teams within companies -- that will be a job requisition out there soon.” Diverse teams truly capable of understanding the impact of bias mitigation will become the norm as corporations develop experiences that benefit consumers from all corners of the globe.

Tags: Ai |

calendarCalendar of Events