True user value of AI Voice Technology

Full Application: Not funded at this stage

There has been a growth in the availability and use of AI Voice technology, e.g. Google Voice / Amazon Alexa, within the service industry, e.g. 1 in 10 British citizens now own such a device (YouGov).

Research shows this is to grow, with more devices becoming AI Voice enabled and an increase in a willingness to access goods and services using AI Voice (ComScore state by 2020, 50% of all online searches will be “voice” searches).

As organisations that offer a variety of services, need to understand the value of this technology for our users/potential users. We want to know what the true user need of AI Voice is and what demand there is and will be for using such technology.

We want to understand who the users would be and from that establish a suitable LA service that would benefit from AI Voice technology that could be used in an Alpha development.

We need to establish what impact, if any, strong accents, regional dialects and speech impediments might have on use of AI Voice – studies have suggested AI programs can discriminate based on its training algorithms (O’Connor, University of Massachusetts).

We would identify:

– User need for this technology

– A service area for an Alpha project

– Impact of accents, dialects and speech impediments on AI Voice

– What’s possible technically within the restraints of LA budgets and development skill sets

– Financial benefits to local government organisations for meeting service demands through AI Voice

The project will take an Agile approach to delivery. Led by Deliver Manager a from Newcastle Council’s Digital First team and equivalent resource support from partner councils. For user research we will commission external specialist user researchers to conduct research to establish true user needs of AI Voice tech. We will work closely with any external resource to learn new skills/techniques. Deliverable goals throughout will be established, MoSCoW’d and prioritised in sprint kick offs. Stand ups will be used to monitor progress and identify blockers. The project will be over a series of sprints (see below), that will end with show and tells that can be recorded to share with full group. Sprints will end with retrospectives to evaluate successes and identify improvements. To assist in communication between the project team we will use tools such as Microsoft Teams, e.g. planner and Skype.

21 Dec18, Kick off meeting / Agreement with project partners

18 Jan 19, Info gathering & data analysis from project councils, Baseline data gathering complete

25 Jan 19, Core team established, Team complete GDS Academy training, Recruit external resource i.e. user researcher(s), Team ready to start Discovery activities

15 Feb 19, Build hypotheses for user research regarding use of Voice AI  and hypothesis regarding services that could use AI Voice, Plan user research

25 Feb 19, Carry out UR, Analysis of inhouse skills/knowledge for AI Voice tech, at project councils and light touch at wider LA level, Research Report, Analysis report on technology

1 Mar 19, Analysis of UR of “service that could use AI Voice” / “use of AI Voice”, Analysis of tech available, Analysis report of UR, Analysis report of Technology

8 Mar 19, “Paper” process designed service, Prototype process developed

15 Mar 19, User testing and analysis of “Paper” process

25 Mar 19, Write up and share the project findings: Business case, User research report, Recommendation report

Despite much work to improve our digital offerings, e.g. promotion of online portals, improved digital forms and introduction of chat bots, phone remains no.1 choice for users to interact with the council. As the SOCITM study estimates this, after face to face, is the most expensive transaction type (£2.83 per transaction compared to £0.15 per online transaction). Digital services are on the rise, but the use of them is not at the rate we would like. A small piece of user research at Newcastle Council revealed 37% of customers calling our contact centre to request/use a service hadn’t used the available online offering due to a skill issue. Some users identified age and/or disability as the reason behind this.

We would like to know if by investing in AI Voice we could assist moving this difficult group to engage with services digitally, and also help them participate in a digital world that by using text based devices is not accessible to them. Disabilities aren’t just physical, they are also be situational. This can cause people to revert to phone to conduct a transaction they may previously have done using physical online e.g. a mother unable to use her device whilst holding their baby, or a driver in their car on a long journey. Within Newcastle’s user research, 25% of customers chose to use the phone, some of this group cited an inability to use their hands at that moment as the reason they chose to phone – i.e. they were out walking the dog. Whilst identifying a service that could benefit from the use of AI Voice, we would want to redesign the service around digital technology, rather than “bolt on” a digital offering to a possibly failing process. By focusing on user need, taking a lean approach to the process design, we can eliminate costly, waste driven processes. The project will also undertake an analysis of technology stacks and skills available with ICT departments of the project collaborators.

We have 4 project partners, and can utilise resource from authorities who have not yet signed the digital declaration. Our group represents city, borough and non-metropolitan councils, so our user base is wide and varied so findings will be relevant across most local authorities. Collaboration will primarily be regarding user research, and findings will be shared with each other through Microsoft Teams. Show and tells will be recorded so they can be shared with other LAs – we will also actively receive feedback from those LAs outside of the project group and allow this to inform any gaps in our research. Our research hypothesis will come from gathering financial, quantitative and qualitative data from project partners. We will also capture information from other government organisations or local authorities, as well as the charity/volunteer sector. We will approach private service organisations to understand how they value AI Voice and their future plans so we can learn any lessons. 

Business Case 

This discovery piece will deliver a business case for the AI Voice technology, relevant not only to the project authorities but to all local government. It will determine the benefit of investing in this type of technology for both the user and the authority. It will also identify skills and or technology gaps that local authorities may have so they can understand this before undertaking similar alpha projects to the one that this discovery piece may (or may not) recommend. 

User Research Report 

Throughout the project we will publish findings of our user research, this will be done through a series of channels, including blogs with a final piece that will take the form of a final summary report. User research will be done using tried and tested GDS templates to allow consistent and comparable outcomes. 

Summary report and recommendations 

The final report will offer a recommendation on the value of AI Voice technology, and if suitable, a service area that would benefit from the use of this technology in an Alpha project. 

The report will also include a review of skills and technology stack required to build any solution, which would identify potential blockers and ways to overcome them. 

Redesigned service map 

If a suitable service is identified during discovery, a light touch redesigned service map incorporating the technology will be developed. 

There are two pieces of user research required: 

  • Understand the user need of AI Voice technology 
  • Identify a service area that users and LAs would benefit from being led through AI Voice technology  

User research will primarily focus on adults who engage with the council via face to face, phone or digitally. 

User research will be done over the boroughs of the project partners, and will be designed and conducted by a user research team that the applied funding will support. 

Both inductive and deductive research will be used to understand what the user need of those wishing to engage with a local authority is, and whether this could be met using AI Voice technology and the service(s) that could benefit. 

Objectives of the research 

  • Understand how our user base views AI Voice technology 
  • Are they willing to engage with it, what do they view as the benefits, what do they view as being potential issues 
  • How much of our user base currently have, or are likely to soon, purchase AI Voice technology e.g. Google Voice or Amazon Alexa 
  • Which services would a user be comfortable engaging with Voice AI to achieve i.e. transactional, enquiry, social service etc.
  • Access to GDS user research labs and research expertise 
  • Access to the GDS training courses, in particular user research at Longbenton DWP/HMRC, the closest for Newcastle and Northumberland, whilst within reasonable distance for Doncaster 
  • MHCLG support and guidance where needed 
  • Assistance in sharing our findings to a wider LA audience