Vodafone and the University of Oxford tackle bias bots, AI, Analytics & Automation

Film makers have lengthy embraced the idea of Artificial Intelligence (AI) robots mimicking human minds, and not at all times with a cheerful ending. Vodafone is working with the University of Oxford to keep away from such an end result. While the alternatives for innovation in AI are immense, the firm can also be guarding in opposition to any pitfalls that come up as the new know-how matures.

Increasingly, Vodafone is popping to AI to supply prospects extra personalised merchandise and providers, prolong community capability the place it’s most wanted, and to resolve buyer enquiries extra rapidly.

But the unintended consequence of such personalisation and enterprise determination making is the threat of algorithm bias. Data used to coach an AI bot might comprise human favouritism or intolerance – like a passion for a sure handset model – or information units that aren’t consultant of all Vodafone’s 350 million cellular, mounted broadband and TV prospects.

A analysis collaboration framework has been initiated and established by Dr Jindong Hou at Vodafone and Professor Thomas Lukasiewicz, Dr Oliver Sampson, Associate Professor Ani Calinescu and Doctoral Student Tatiana Botskina at The Department of Computer Science, University of Oxford in the final three years. Thanks to this initiative, Vodafone is working with the University of Oxford to embed moral AI all through the buyer journey, to make sure honest, clear, explainable, and accountable AI.

The Age of Big Data

Dr Jindong Hou, Lead Research & Innovation Architect at Vodafone, defined: “In the age of large information, an organization that harnesses the energy of AI can have a aggressive edge. However, as AI grows in utilization and affect throughout geographies and industries, we’ve got the accountability to think about the way it impacts our prospects, our workers, and wider society.”

The march of the AI bots exhibits no signal of abating. Most buyer on-line interactions contain rising applied sciences akin to machine studying (ML), AI-driven chatbots and cellular messaging. My Vodafone App, which provides prospects personalised providers and rewards, is now accessible throughout 17 markets, and TOBi – Vodafone’s chatbot – is studying new tips all the time akin to proactively conducting community diagnostics to assist present a greater cellular sign to prospects.

These are examples of AI-driven advantages. Jindong harassed that Vodafone pays specific consideration to making sure its AI applied sciences are designed to respect the privateness and safety of the finish person’s information and their related basic rights. “The buyer information Vodafone makes use of is pseudonymised and permissioned,” he mentioned.

Ethical AI by Design

Now, Vodafone needs to go a step additional and embed ‘moral AI by design’ at the begin of each on-line buyer journey, whether or not that’s ordering a brand new system, or querying a invoice. The analysis collaboration venture with the University of Oxford will assist form honest, clear, explainable, and accountable AI.

“We are the potential of deep studying fashions to know hidden patterns and dependencies. These will permit us to higher analyse and achieve insights on advanced enterprise processes, akin to buyer journeys, to detect and take away human bias,” continued Jindong.

A proposal written by associates of the University of Oxford reinforces this view. It describes the advantages of combining attention-based machine studying fashions (which break down a posh process akin to a buyer trying to find, and shopping for, a product) with causal inference (understanding the trigger of a choice from the information accessible). Getting to grips with it will assist Vodafone additional perceive what frustrates prospects and what delights them.

Attention-based language fashions assist AI to deal with the most necessary components of language, a picture, or some other sample that’s being analysed. This helps to enhance the output, which is likely to be a translated sentence or a chatbot response. It has been used to nice impact in healthcare, serving to to foretell the greatest therapy for a affected person.

Free of Harmful Bias

Using a mix of attention-based machine studying fashions to search out patterns, and then people to assist interpret or infer the causes of these patterns, Vodafone will be capable of perceive its prospects higher. The firm can determine relations between completely different occasions inside a buyer journey and determine and management for any potential biases.

Jindong continued: “Many of the points that come up in the context of AI and human rights are usually not novel, however are exacerbated by the scale, proliferation, and real-life affect that AI facilitates. Because of this, the potential of AI to each strengthen and diminish human rights is far higher than in earlier waves of technological growth.

“Wherever doable our AI-driven selections are honest and free of any dangerous bias and respectful of information privateness and safety.”

Speed-up Boring Processes

Vodafone can also be trying to make use of AI to make its workers lives simpler by making use of methods to scale back mundane duties that release time to innovate and higher serve prospects. One such space that Jindong highlighted is Natural Language Processing (NLP) – a subset of AI through which machines make sense of human language.

“Let’s be sincere, most of us discover it tedious to jot down minutes both throughout or after conferences. Therefore, we developed an AI-based answer referred to as Meeting Minutes Automation, which precisely targets this downside by totally automating the minute taking process.

“The bot is full of helpful issues, together with having the ability to convert speech to textual content, print solely the phrases of sure audio system, detect motion factors and extract key phrases. It is just not solely good; it saves time, cash and speeds-up boring processes.”

AI alone is just not the reply to making sure a clean buyer journey. Vodafone has invested closely in changing legacy methods, which don’t discuss to one another, with people who take benefit of our scale to offer prospects with a extra constant service throughout borders.  For instance, Vodafone is bettering its community planning decision-making by utilizing software program that may course of and analyse as much as eight billion factors of information daily from throughout its cellular community in 11 European international locations.

In parallel, Jindong and the staff are eliminating any unintended bias in Vodafone’s decision-making. They are working onerous to make sure AI horrors performed out on the silver display screen are usually not replicated on the smartphone, PC, or small display screen.


Recommended For You