Blog:Natural Language Processing ala Google AI & ML
Let's level set. Used right and driven with compassion, anti-racist awareness and action, and an intent for redesiging oppressive systems, technology can greatly supplement our effort to do good work efficiently and at scale. By extension, I'm continously looking for, evaluating and learning from the various tools, services and best practices that are in the external and internal ecosystem of any team and company I'm on for helping solve real-world injustices and problems.
Used right and driven with compassion, anti-racist awareness and action, and an intent for redesiging oppressive systems, technology can greatly supplement our effort to do good work efficiently and at scale. PS - Technology is inherently political
My team is at a point where we're outgrowing our startup's machine learning implementation for how we're reliably analyzing data and training our model. Naturally, I've been researching to see what tools may be available that allow me and my team to scale our product with confidence while not breaking the bank. Enter Google's AI & ML product suite.
Tech is not neutral. Leverage your privilege and power to redesign and undo flawed tech and product ecosystems.
Google's AI & ML Product Suite
Simply put: the platform and products are reliable and scalable whether you're an individual, small shop or enterprise level company. There is a learning curve to it, like anything else, but the team behind the product suite has sure come a long way in making the platform more easily accessible since the earlier days. The suite offers some good NLP options to leverage such as entity analysis, content classification, sentiment analysis and more; effectively, helping you give structure and meaning to your data.
NLP Sentiment Analysis Example
During a Sunday morning, I extended my existing node API service on my site (this one) to be able to run NLP sentiment analysis—using Google's platform—on arbritary text that I (or you) feed it. For clarity, extending my API to accomplish this was not entirely related to Google's AI & ML platform, this effort is merely a part of any fullstack development solution, however, Google's API client tools sure made connecting with their AI & ML platform nearly friction-less when the time came to wire things up from my API server to their platform. Side note: I did come across a couple of
incomplete areas of documentation that could have saved a few minutes of headscratching, but it wasn't that big of a deal. I'll see about submitting a pull-request with the Google Language team to address this.
Once the API was extended, I created some custom UI components ala React and my design system that led me to this quick-and-dirty proof-of-concept that shows how both texts and individual parts of texts (e.g. sentences) are analyzed:
For those interested, I'll write a separate post showcasing how to train your own custom model with Google AI & ML. I'll plan to go into the code weeds of the approach if it helps folks more easily learn and get started in giving things a try themselves.
Where to go from here
The upshot: As a part of my team's initiative to scale and continuously improve our platform, product and services, I believe Google's AI & ML product suite is going to be a top contendor in what we evaluate this year to help us reach our goals and level-up our product for our customers.
I am optimistic about the challenges ahead of us, along with the rewards, with the advent of AI/ML. I am also realistically concerned if we don't collectively take proactive steps to prepare for the far reaching impacts AI/ML has on society, such as on individual relevance; economic, racial and social (in)justices, and workforce displacement.
Whether you're trying to understand customer feedback to find actionable product and UX insights, or you want to do some social good, or accomplish something with ML for a personal research project, or have some school project revolving around AI, I'd recommend giving Google's AI & ML product suite a whirl. I'm happy to jump on a call with anyone who hits a snag during setup and would like some high-level guidance on ways to configure things.
Let's do some good.
- A11Y: Inclusive Design
- Bias in Artificial Intelligence Confirmed
- Discriminating Systems: Gender, Race, and Power in AI
I operate from a place of compassion, possibility and imagination. My work and efforts share a common goal: create a better, sustainable and equitable world by building inclusive communities, products & experiences.