Services are said to use the “Teach and Test” methodology to ensure the quality of AI systems.
First comes the hype, then comes the products, then comes the realisation that no one actually knows what they’re doing.
The traditional song and dance around a new technology, in this case Artificial Intelligence, typically either sees the tech fail to reach its potential, or bumble along for a while, never really reaching the promised lofty heights.
Artificial Intelligence has already seen it’s own false dawn, but this time there’s a much more advanced tech community there to support and nurture its advancement.
So, whilst every person and their dog throws AI at their products, Accenture is aiming to help businesses in validating the safety, reliability, and transparency of their AI systems with a new testing service.
The Accenture AI Testing Services plays off a “Teach and Test” methodology, providing businesses with the ability to build, monitor, and measure reliable AI systems within their own infrastructure, or in the cloud.
The “Teach” phase looks at the choice of data, models and algorithms that are being used to to train machine learning, statistically evaluating different models so that the best performing one can then be deployed into production.
It also promises to avoid gender, ethnic, and other biases, as well as ethical and compliance risks.
The “Test” phase sees the AI system outputs compared to KPIs and assessed as to whether or not the system can explain how a decision or outcome was determined.
“The adoption of AI is accelerating as businesses see it’s transformational value to power new innovations and growth,” said Bhaskar Ghosh, group chief executive, Accenture Technology Services. “As organisations embrace AI, it is critical to find better ways to train and sustain these systems – securely and with quality – to avoid adverse effects on business performance, brand reputation, compliance and humans.”
The “Teach and Test” methodology has already been used by Accenture in order to train a conversational virtual agent for a financial services company’s website, the idea being that it would be able to engage in conversation before referring to a human if necessary. The outcome was that the agent was trained 80% faster than previously possible, and achieved an 85% accuracy rate on customer recommendations, according to Accenture.
“Testing AI systems presents a completely new set of challenges. While traditional application testing is deterministic, with a finite number of scenarios that can be defined in advance, AI systems require a limitless approach to testing,” said Kishore Durg, senior managing director, Growth and Strategy and Global Testing Services Lead for Accenture.
“There is also a need for new capabilities for evaluating data and learning models, choosing algorithms, and monitoring for bias and ethical and regulatory compliance. Accenture’s “Teach and Test” methodology takes all of this into consideration to help companies develop and validate AI systems with confidence.”