The evolution of no-code AI

• Read original article here

What – and where – is no-code AI?

A code-free system that empowers companies using artificial intelligence (AI) to perform various activities such as, but not limited to, data classification and analysis, “No-code” solutions enable users to create application functionality without writing programming code

When placed into the context of AI, machine learning (ML), and data science, the overall objective of no-code AI is to democratize data analytics and data insights.

The power of no-code AI lies within its potential to enable less “technical” folks – product managers, sales, store owners – to have a more intimate relationship and interaction with their data. By leveraging no-code AI, those disinclined toward the complexities of coding can directly explore, analyze, test hypotheses, and make predictions out of their valuable data. All without having to dive deep into data science, or rely on a team of data scientists to execute tasks that achieve their goals.

I think about how J.A.R.V.I.S. from the Marvel movies, functions as Tony Stark’s AI assistant, running and taking care of all the internal systems of Stark’s buildings and the Iron Man suits, but we are a long way from that in the real world.

Functionally, no-code AI should contain a few building blocks:

At the foundation, it needs the capability to conduct many common ML/AI tasks such as data ingestion, data cleansing, data QA, feature extraction, training models with various ML techniques, parameter search, and model evaluation.

Building on top of the foundation, there needs to be a business logic layer that enables no-code AI to assist the users to solve specific types of business problems in addition to simple generic ML tasks.

It should also have an intelligent presentation layer that is able to automatically create the most appropriate visualization to present the information to the users.

To bridge between these AI capabilities and the less technical users, no-code AI needs powerful natural language processing/understanding/generation (NLP/NLU/NLG) capabilities to interact with the user using plain English. This is necessary to understand the user’s questions and invoke the right analysis and/or even directly translate the request into executable codes. It needs to go beyond Alexa and Siri, or even the existing task-oriented conversational AI/chatbot solutions that can look up weather conditions or book flights. Those solutions need to predefine the types of questions that can be asked, the individual pieces of information relevant to those questions, and the actions to take based on the questions. Ideal no-code AI should allow users to ask any questions pertaining to the data in the database.

Until we have true AI that conduct analysis independently, we won’t quite be at the stage in which no-code AI can free users from needing basic knowledge surrounding the problems the AI needs to address. For example, asking no-code AI to predict who is going to attrite is fairly straightforward, but the definition for attrition can vary in business settings. It’s possible a bank customer has nearly stopped using their bank account, yet the account remains open. In practice, this should be considered as attrition, but it is often not marked in the bank’s system as closed. Having no-code AI simply use a closed-account flag in the data as the intended target definition for predicting attrition would yield, in this instance, a less desirable outcome for the business.

The promise of no-code AI and its applications is significant and will continue to power data analytics and insights. By continuing to evolve and understand its potential, users will be able to make the most of this rapidly evolving technology – and stay ahead of potential risks.

Kevin Chen is the Chief Data Scientist of Experian DataLabs in North America. Experian DataLabs helps businesses solve strategic marketing and risk-management problems through an advanced data analysis process and research and development. DataLabs helps deliver innovative new data sources with an emphasis on financial services, telecommunications and healthcare.