Generative AI is Here to Stay
As we look back at 2023, one thing is clear that Generative AI has entered in many aspects of our personal and professional lives. The recent advances in many Large Language Models have made AI practically accessible and useful for many tasks, from getting more contextualized search results to automating sales processing using natural language instructions. Though 2023 was a breakout year for Generative AI, a mid-April 2023 McKinsey [1] survey findings show that, despite Generative AI’s nascent public availability, experimentation with the tools is already relatively common, and respondents expect the new capabilities to transform their industries. Generative AI has captured interest across the business population: individuals across regions, industries, and seniority levels are using Generative AI for work and outside of work. Seventy-nine percent of all respondents say they’ve had at least some exposure to GenerativeAI, either for work or outside of work, and 22 percent say they are regularly using it in their own work. While reported use is quite similar across seniority levels, it is highest among respondents working in the technology sector and those in North America.
Software Engineering/IT is the Tip-of-the-Spear
It makes perfect sense that the technology community in enterprises is the tip-of-the-spear when it comes to AI. They are driving AI in both dimensions – making it more scalable, reliable, and affordable to use and using it to simplify and accelerate their tasks. There is also a strong precedence of internal engineering or IT organizations “eating their own cooking” before big technology changes became industry norms. For example, Cloud migration picked up when Amazon’s engineering organization offered the product they had been using internally to their enterprise customers.
The advances in Generative AI have enabled its incorporation into various aspects of the software development cycle, brainstorming ideas, gathering requirements or even writing code, assisting in creation of UI elements, handling code migration or updates, etc. [2] [3]
AI Augmented Software Development Lifecycle
Path to AI Benefits is full of Challenges
Though AI is already helping in writing code snippets and automating dev-ops tasks [4], there are many challenges in harnessing full potential of AI in the entire software engineering process:
- Training methodology: LLM models are only as good as the training method and data set used. The training methodology needs to go beyond just the raw datasets. A vast majority of human learning comes from visual information, particularly when humans have to interact with something using that information. Computer Vision assisted learning can be a good solution to achieve this in AI provided it learns from visual output and not from the programming structure that produced that output.
- Cognitive learning: For AI to be effective in Requirements, Design and Testing phase of software engineering, it needs to understand how humans use a software application. Writing logic and using a software are two entirely different things. For building better software applications for human consumptions, AI needs to know how humans use software applications.
- Probabilistic vs confirmative: AI is inherently probabilistic when it comes to its outcome. We will need a human-in-the-loop when the outcome deviates from what is expected. This is a very important consideration when consuming AI output as it is. Many technology executives from large enterprises echoed the same concern – we know our teams are using AI tools to generate code but who is making sure that code is error free?[5]
- Checks and balances: When vendors introduce new, intelligent features—often with little fanfare—they are also introducing models that could interact with data in the user’s system to create unexpected risks, including giving rise to hidden vulnerabilities that hackers might exploit. The implication is that leaders who believe they are in the clear if their organization has not purchased or built AI systems, or is only experimenting with their deployment, could well be mistaken.
The Saralam Way to AI Benefits
At Saralam, we are taking a very thoughtful approach to harnessing the power of AI. We are ready to help software engineering organizations to demonstrate reliable, frictionless, and cost-effective approach to benefit from AI today. Saralam’s No Code Testing Service addresses relatively lower risk but complex to automate area of test design and development. Our custom trained models using computer vision and guided exploration can implement test cases quickly for any software application and maintain those as the application evolves. Our services approach makes sure that there is a human in the loop to verify that the test cases are legitimate and accurate.
Saralam No Code Testing provides a quick and credible way for engineering organizations to demonstrate AI benefits to the rest of the company and be a champion in expanding those in other functions.
References
[1] Chui 8/01 [Chui, Michael. “The State of AI in 2023: Generative AI’s Breakout Year | McKinsey,” 08/01/2023. https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai-in-2023-generative-ais-breakout-year .]
[2] Karaci Deniz et. al 23 [Begum Karaci Deniz, Chandra Gnanasambandam, Martin Harrysson, Alharith Hussin, and Shivam Srivastava. “Unleash Developer Productivity with Generative AI | McKinsey,” 06/27/2023. https://www.mckinsey.com/capabilities/mckinsey-digital/our-insights/unleashing-developer-productivity-with-generative-ai]
Panetta 23 [Panetta, Kasey. “Set Up Now for AI to Augment Software Development.” Gartner, 09/21/2023. https://www.gartner.com/en/articles/set-up-now-for-ai-to-augment-software-development.]
[3] Gartner 2022 – Market Guide to AI Augmented Software: [By 2027, 70% of professional developers will use AI-powered coding tools, up from less than 10% today; 80% of the enterprises will have integrated Artificial Intelligence (AI)-augmented testing tools into their software engineering toolchain, which is a significant increase from 10% in 2022.]
[4] GitHub reported a million developers have used Copilot and 20,000 organizations have adopted it.
