Applause Launches AI Training and Testing Solution

Today, we announced a new solution for the training and testing of AI. Leveraging our global community, this new end-to-end solution delivers the widest possible range of training inputs, and then tests the results across every possible device, location and circumstance to identify issues and provide actionable user feedback in real time. This enables brands to design, deliver and maintain high-quality AI experiences for all their customers.

Training a machine to learn, think, act and respond like a human takes massive amounts of data inputs across countless potential scenarios. That data needs to come from a broad variety of people with different backgrounds, experiences, ways of thinking and behaving, and more, in order to capture the full spectrum of human behavior as accurately as possible. Trying to source data diverse enough to train algorithms accurately, at scale, and quickly enough to continually train and test, is a tremendous challenge.

Not only is there the internal challenges of sourcing and testing data, but there are outside pressures as well. Although AI-powered experiences are relatively new, customers have high expectations and are disappointed when AI fails to deliver on its promise to make life easier and more efficient. Whether a voice-enabled device doesn’t understand a request, a recommendation engine displays irrelevant suggestions or a chatbot responds with generic answers – sub-par AI experiences can frustrate and alienate users instead of delighting and engaging them.

Applause provides a different approach to AI, using the community to test AI-driven user experiences throughout the development process to validate that algorithms are producing accurate, human-like and truly useful results. The community can be deployed to source any dataset required, whether it’s text, images, speech, handwriting, biometrics or anything else. This data comes from real people across numerous countries, ages, genders, races, cultures, political affiliations, ideologies, socioeconomic and education levels, and more. This diverse training data results in a more broadly representative and unbiased output than if the data were sourced from a smaller group.

The diversity of the community also ensures brands are getting feedback across a wide variety of people quickly and at scale, so brands can iterate on machine models and produce the results that users expect and demand.

The new AI training and testing solution operates across five unique types of AI engagements:

  1. Voice: Source utterances to train voice-enabled devices, and test those devices to ensure they understand and respond accurately.
  2. OCR (Optimized Character Recognition): Provide documents and corresponding text to train algorithms to recognize text, and compare printed docs and the recognized text for accuracy.
  3. Image Recognition: Deliver photos taken of predefined objects and locations, and ensure objects are being recognized and identified correctly.
  4. Biometrics: Source biometric inputs like faces and fingerprints, and test whether those inputs result in an experience that’s easy to use and actually works
  5. Chatbots: Give sample questions and varying intents for chatbots to answer, and interact with chatbots to ensure they understand and respond accurately in a human-like way.


Learn more about the solution on our new AI training and testing webpage.

Webinars

Sourcing Training Data for AI Applications

Once you’ve made the decision to leverage AI and/or machine learning, now you need to figure out how you will source the training data that is necessary for a fully functioning algorithm.

Watch now
Want to see more like this?
Dan Haggerty
Writer
Reading time: 5 min

How to Assess Testing Resource Allocation

If you can’t measure the value of your efforts, you can’t explain or even justify your testing investment

Using Questionable Datasets to Train AI Could Come With High Costs

As companies look to capitalize on AI development, they must stay mindful of how they source training data — AI algorithms developed from private or non-consensual data may cost businesses in the long run.

Why Payment Testing is a Constant for Media Companies

Simulated transactions and pre-production testing won’t ensure friction-free subscriber payment flows

How Mobile Network Operators Can Leverage e- and iSIMs

We explain what e- and iSIMs are, what they mean for the industry and how MNOs and MVNOs can leverage them to their advantage.

Jumpstart Your Software Testing Education

Testers have a variety of options to upskill and grow professionally

The Future of Generative AI: An Interview with ChatGPT

We ask ChatGPT about where it sees itself in the future, what needs to happen for it to get there and how Applause can help.