Discriminating AI: The pros and cons of artificial intelligence

lines

The surge in Digital Transformation

Never has there been an impending rush where organisations are falling over to digitalise themselves with existing and emerging technologies, to change the way they operate, connect with customers, automate work and use data. Artificial Intelligence emerges at the forefront of this digital evolution and is woven into the fabric of human existence, with future predictions to be the source of all aspects of decision making.

Sounds amazing – everyone is a winner, right?

Well, with 7.6 billion people on the planet, the scale of diversity and ambition to create products, experiences and services that will please everyone, is always going to be a challenge; however, when it benefits a handful, there is a huge problem. A recent University study (1) titled ‘Discriminating Systems’ has addressed deeper issues related to the workplace cultures, power asymmetries, harassment, exclusionary hiring practices, unfair compensation, gender inequality and tokenization in today’s AI industry.

AI applications use deep learning through a series of algorithms to find patterns in data, which is used to solve problems and provide solutions. Some organisations use AI algorithms for suggested outcomes based on location, gender, race and other statistics, which automatically exclude people based on those very same factors. There are security systems and banking organisations adopting AI that uses algorithms to single out people from certain backgrounds or provide selected financial products to maximise profit – resulting in predatory behaviour, even if this was unintended by the organisation. Studies have highlighted how governments, legal bodies and judicial systems have used AI technology to assist them with their decision making, resulting in outcomes based on algorithms that appear to be biased based on gender, race, social status and origins.

fatos-bytyqi-Agx5_TLsIf4-unsplash-450x550

So where did it all go wrong?

AI isn’t new – emerging in the 1950s in its simplest form, it has now evolved into an almost living, breathing and dynamic omnipresent entity.
Studies have emerged showing that bias starts with people who programme these intelligent systems through either training data or existing prejudices. Global organisations are now dealing with ramifications for not having a diversified workforce to build inclusive systems. These studies highlighted that more than 80% of AI professors are men, with less than 18% women. They reveal that large-scale AI systems are widely developed by a handful of technology organisations that tend to be dominated by white males, affluent and technically orientated (2), with a small representation from women and diversified backgrounds.
However, let’s not jump the gun and attribute complete blame, as not all standard practise and deep learning was designed to be biased. This rapidly evolving technology requires constantly training, re-training and programming to act in a certain way; and each of these steps involving human influence presents risks of bias creeping in. The lack of transparency and accountability is resulting in badly behaved systems.

Sounds serious – why isn’t more being done about it?

On the 8th April 2019, the European Union, supported by expert views, published new guidelines on developing ethical AI, covering 3 main areas and suggesting that trustworthy AI should be:

  • lawful – respecting all applicable laws and regulations
  • ethical – respecting ethical principles and values
  • robust – both from a technical perspective while taking into account its social environment.

Microsoft recognised this earlier on in 2016; and, led by Tim O’Brien, operate a cross-functional committee addressing AI and ethics in engineering and research. They run a programme called ‘Inclusive Design’, which is a methodology, borne out of digital environments, that enables and draws on the full range of human diversity.
IBM recognised the issue and has been creating awareness through research, fearing an increase in the number of biased AI systems in the next 5 years. They actively work towards mitigating bias in AI systems to build trust and transparency; develop learning systems that recognise impartial, egalitarian views; and learn from common human values in decision making.
The Focus is to make organisations more accountable.

So, is all AI bad?

Far from it, and I don’t think we can ever blame AI. That’s like saying ‘my oven is terrible and always burns my food’, when in fact you are responsible for setting temperature, timers, recipes, etc., and following instructions – which would be fine, I guess, if you were the only one eating that food. However, organisations use AI to aid in their decision making and to improve processes, improving customer experience, which affects several people, depending on the organisation’s size. These intelligent systems aren’t human; however, they are designed to understand and cognise human behaviour and are only as good as the people who design them and their training data.
Let’s not forget the countless benefits that the AI industry has brought us:

  • RPA helps remove routine and repetitive tasks, eliminates human error, and is moving closer towards a complete dissolution to back-office activities, thereby supporting human workforce to focus on customer-centric tasks.
    We are all touched by Virtual Assistance making our lives easier in one way or the other, be it Siri, Alexa, or technology that allows you to remotely conserve energy and fuel at home.
  • Machine Learning creates a better online shopping experience; and in the world of media like Netflix, learns from your viewing history and identifies your viewing behaviour to customise your future watch list. Machine learning in devices uses predictive behaviour to assist with scheduling, planning and making lives more efficient.
  • Contact centres use Natural Language Call Steering, Intelligent IVRs, Chatbots, etc., to increase the customer experience and allow their human staff to work on more complex issues.
  • Augmented Reality & Virtual Reality revolutionised the travel industry by helping with planning, navigation, and sight-seeing. And, think about the reduction in carbon footprint in retail, where AR & VR have reduced the need for physical space, travel and production, by experiencing products and services remotely and creating opportunities to manufacture on demand.
VR-250x445

So, what can organisations do to find the right solutions?

It’s easy to be attracted to everything that’s shiny and new and buy into what may seem like the perfect solution for helping your organisations achieve their digital evolution. However, before you jump in, ensure you do your research and/or engage with specialists who can help you make the right choices:

  • Establish your requirements and develop what is right for your organisation and customers
  • Find the right products to meet your requirements
  • Develop your procurement process by asking the right questions and conducting the right research to facilitate making the right decisions.
  • Once you find the right solution, ensure you have the right people to help you implement it
  • Test your solution design to ensure that it’s ethical and meets inclusion and diversity
  • Increase your transparency to build trust

Ember works with leading organisations to identify, procure and implement their digital solutions. Recently Ember worked with one of the largest hospitality apps in the world to help them build out their future capabilities, map out emerging technology that is right for their business and develop their operational thinking towards changing customer expectations and realities.

About the Author: Neil Chitre

With over 18 years’ experience in the contact centre industry, Neil has led the operational delivery and digital transformation of contact centres across Europe and Asia. He has experience in multiple sectors including FMCG/CPG, Retail, Automotive, Healthcare and Financial Services, with further skills in outsourcing, knowledge management, social media and demand reduction through automation. Recently, Neil has worked on advising the most recognisable accommodation site in the world on their future technology and is currently working with one of the largest retailers in the world, to help them to shape their global operational framework to support their CX transformation.

Neil-BandW-200x200