Accelerating AI Adoption
Jaime Tatis, Vice-president Data Strategy & Enablement, TELUS & Alexandre Guilbault, Director of AI Accelerator, TELUS
Jaime Tatis, Vice-president Data Strategy & Enablement, TELUS & Alexandre Guilbault, Director of AI Accelerator, TELUS
TELUS’ journey through digitization and analytics leadership over the years has been leveraging cutting-edge AI techniques and technologies across all verticals to drive better user experiences and products. At its core, AI adoption is a key component of improving customer experiences and one that Jaime Tatis, Vice-president of Data Strategy & Enablement and Alexandre Guilbault, Director of AI Accelerator at TELUS, both know very well. Read more below for their take on the future of AI.
It's likely that we're witnessing one of the most prolific periods for AI with major releases of impressive text and image generators such as ChatGPT, Bard and Dall-E. These generative AI models have been making headlines – sometimes writing the news articles themselves – but these commercial applications are just the tip of the iceberg when it comes to AI’s true potential.
Like any industrial revolution, an imminent transformation is ongoing, and although frightening to some, the benefits of responsibly operated AI systems are enormous for industries seeking new and better ways of working. Not only is AI adoption accelerating businesses forward, but it is also helping to create smarter investments that will support customer experiences and reduce the strain on individual employees. For instance, TELUS leverages Google's CCAI Insights tool to automate the process of capturing, transcribing and analyzing customer interaction, leveraging these insights to empower agents and provide leading improved experiences. TELUS is also piloting and doing first deployments of use cases powered by generative AI technology to improve operational efficiency.
Additionally, AI technologies offer industries the potential to leverage large amounts of data to optimize decision making, increase efficiency and productivity, and reduce repetitive tasks so that employees can support customers directly with more complex challenges. For example, by joining massive network performance KPIs, we have been adopting a proactive approach to network monitoring and issue resolution. We have been addressing network problems swiftly, sometimes even before they occur. This approach not only saves employees valuable time but also improves network reliability, leading to a better customer experience.
Collaboration as a pillar for success
A successful AI transformation needs proper alignment between the business teams and the data experts. The real tangible value cannot be generated without an interdisciplinary collaboration.
One effective approach is to organize stakeholders from multiple teams in pods or squads with common goals. The data experts then need to adopt a strategic consulting approach to problem-solving, prioritizing the business outcomes of their work over their technology metrics.
In a competitive and fast-moving environment, having even a partial understanding today can lead to significantly greater benefits than waiting for a complete understanding at a later time and missing a timely opportunity.
Accepting a new mindset
Although the benefits are high, the rapid pace of technological change has brought about significant challenges and risks for companies looking to embrace AI transformation. Implementing AI systems requires access to high-quality and relevant data, considerable computing power, and the ability to address factors outside of technology itself, such as security and privacy compliance, end-user engagement, and trust in the insights generated. Furthermore, a willingness to change and transform work is critical for success, and the failure to consider these factors can result in projects that do not produce meaningful financial outcomes.
In order to reduce these risks, companies need to adopt an agile, explorative and adaptable mindset. TELUS has been using a Multi-Armed Bandit approach to structure our prioritizations. Having a risk-weighted metric to compare opportunities helps us better select the right focus. Doing so, we are creating an effective "fast-experimental" culture where teams are encouraged to recognize when a project is unlikely to deliver the full expected outcome and when time has come to move onto the next best opportunity. With an incentive to constantly explore for alternatives, each squad can reduce the time spent on projects that may not produce a meaningful outcome and maximize time spent on successful projects.
It is also important to track the right performance metrics and monetize at each stage of a project - not only once completed. Starting with insights in the form of consolidated data or statistics can be a great way to prove value early on, before moving up to more powerful tools that AI technologies are providing.
Prioritizing digital safety
With the switch to AI, companies require a strong foundation in data management, collaboration across communities of data scientists and a relentless focus on prioritization on common goals. Equally important are processes required to efficiently ingest and then manage our data assets into a cloud environment while following high privacy and security standards to protect customers and employees.
All of these technologies can come with risks. Without question, a safe, humanized AI experience comes through the intentional consideration of the impact that the use of data has on people and ensuring that there is a benefit every step of the way.
The TELUS approach to responsible AI builds upon our Customers First commitment and is structured on the TELUS Trust Model by using data in a way that builds trust by generating value, promoting respect and delivering security. Our process begins with clear accountability for our use of AI, ensuring that the technology is the right fit for the circumstances. From the outset, we are thinking about the particular beneficial outcome we are working towards and keeping fairness and the human impact top of mind. This approach includes planning for how we will explain the model and establish ongoing monitoring to ensure it continues to work as intended. We use robust tools and processes which build upon our foundation of Privacy and Security By Design principles. By considering potential bias and using tools that allow users to explain how AI models arrived at a conclusion, we actively strive for fairness through the responsible use and enablement of data. For example, large language models (LLMs) are trained on billions of human created data sets which contain bias, stereotypes and yes - mistakes. However, with the added layer of human oversight, we can help to potentially mitigate those issues to realize further benefits.
A look ahead
AI solutions are meant to boost productivity and allow us to be more efficient but there they all required oversight in terms of reliability, responsibly generated content and ensuring we continue innovating and creating new ideas, not regurgitating the same ones
.However, this adoption of AI is truly transforming the way companies operate and create value for their customers, while also allowing companies to be more socially responsible. Embracing digital transformation, driving collaboration, prioritizing digital safety and accepting a new mindset are crucial as we move forward in ensuring that AI projects are successful. But we know that by embracing this change, we can tailor offerings and services to help meet specific needs and create a stronger and improved experience for our operations and most importantly – our customers.
Weekly Brief
Read Also
