Towards a Technological Singularity with an Android Smartwatch Product as an Example
Technological singularity
We have seen the point of humanity reaching technological singularity many times in science fiction films: the creator of the technology no longer understands how it works and will develop. Of course, similar processes are happening in the field of software development. Since mid-2023, the concept of AI-augmented software development continues to become increasingly popular.
The development case seems interesting Alter Ego Smartwatch Appswhich considers the use of artificial intelligence in the AI-augmented software development paradigm. Alter Ego is an artificial intelligence tool (a personal assistant with 100+ functions and the “intelligence” of a 25-year-old European person), which was created by a team of engineers using other AI tools.
Android Product: AI Writes AI
The development project ran from July 2023 to March 2024 and involved a distributed IT production team.Slavasoft companies. The software product is an English-language Android app for smartwatch “Alter Ego” (Alter Ego), implementing the functions of a “smart” assistant / digital companion. The release development took place in two stages: MVP and Major. Alter Ego software product for Google Wear OS is available for download from the Google.Market store for free and without geographic restrictions.
The special irony of the project was that the artifacts generated by the AI tool and the code base for the application became the basis of the Alter Ego artificial intelligence – in fact, this is precisely an element of technological singularity: “machines” make other and more complex “machines”.
AI-augmented software development: has the future arrived?
The results obtained from the application of AI tools allow us to draw conclusions about the practices of AI-augmented software development.
1) Generative LLMs allow you to create large-scale software products in a short time: the Alter Ego app has 100+ features and speaks like a 25-year-old adult.
2) Code generation is an iterative process. The engineer's persistence in repeating prompt requests and significant detailing of system requirements are crucial.
3) All application content, documentation, auto tests and test cases are easily and completely executed in the LLMs tool.
Conclusion
The given case confirms the fundamental possibility of successfully using AI tools in creating software project artifacts at all stages of its life cycle – from technical specifications and code to documentation, which made it possible to achieve the highest speed of software development for the team.
And although machines cannot yet create new machines (and start the apocalypse), we see practical examples of creating AI software solutions using previously created AI tools as very effective.
