More accessible low-code and no-code offerings from Amazon Web Services (AWS), Azure, and Google Cloud will allow ordinary people to create their own AI applications as quickly as they can deploy a website. Scientists’ highly technical AI applications can give way to lightweight consumer apps that are much more user-friendly.
Consumer-grade AI context
The “consumerization of IT” has been an ongoing theme in tech circles throughout the 2010s, but as of 2022, most enterprise software offerings remain clunky, inflexible, and highly technical. This paradigm is partly due to too much legacy technology and systems still operating within most government agencies and Fortune 1000 businesses. Creating user-friendly AI is no easy task, and it often gets pushed to the side in favor of other priorities like cost and delivery timing.
Additionally, many smaller companies lack the in-house data-science teams that can customize AI solutions, so they often rely on vendors that offer applications with built-in AI engines instead. However, these vendor solutions might not be as accurate or tailored as models created by in-house experts. The solution is automated machine learning (ML) platforms that allow workers with little experience to build and deploy predictive models. For instance, the US-based company DimensionalMechanics has enabled customers to create detailed AI models simply and efficiently since 2020. The built-in AI, referred to as “the Oracle,” provides support to users throughout the model-building process. The company hopes that people will use various AI applications as part of their daily work routines, similar to Microsoft Office or Google Docs.
Cloud service providers have increasingly implemented add-ons that would make it easier for people to build AI applications. In 2022, AWS announced the CodeWhisperer, an ML-powered service that helps improve developer productivity by providing code recommendations. Developers can write a comment that outlines a specific task in plain English, such as “upload a file to S3,” and CodeWhisperer automatically determines which cloud services and public libraries are best suited for the specified task. The add-on also builds the specific code on the fly and recommends generated code snippets.
Meanwhile, in 2022, Microsoft’s Azure offered a suite of automated AI/ML services that are no- or low-code. An example is their citizen AI program, designed to assist anyone in creating and validating AI applications in a real-world setting. Azure Machine Learning is a graphical user interface (GUI) with automated ML and deployment to batch or real-time endpoints. Microsoft Power Platform provides the toolkits to rapidly build a custom application and workflow that implements ML algorithms. End-business users can now build production-grade ML applications to transform legacy business processes.
These initiatives will continue to target individuals with minimal to no coding experience who want to test AI applications or explore new technologies and process solutions. Businesses can save money on hiring full-time data scientists and engineers and can instead upskill their IT employees. Cloud service providers also benefit by earning more new subscribers by making their interfaces more user-friendly.
Implications of consumer-grade AI
Wider implications of consumer-grade AI may include:
- A growing market for companies that focus on developing no- or low-code AI platforms that can enable customers to create and test applications themselves.
- An macro increase in the rate of digitization of public and private operations.
- Coding may become a less technical skill and may be increasingly automated, enabling a broader range of workers to participate in creating software applications.
- Cloud service providers creating more add-ons that will automate software development, including being able to scan for cybersecurity issues.
- More people opting to self-learn how to code by using automated AI platforms.
- Coding education programs being increasingly adopted (or re-introduced) into middle and high school curriculums, fearing these no- and low-code applications.
Questions to comment on
- If you have used consumer-grade AI applications, how easy were they to use?
- How do you think consumer-grade AI apps will fast-track research and development?