Can ChatGPT Outperform Human Testers? - A Closer Look

 

Website-Can-ChatGPT-Outperform-Human-Testers----A-Closer-Look

Artificial intelligence (AI) has been making waves across multiple industries, and the field of software testing is no exception. ChatGPT, a large language model, has been an ongoing subject of concern among testers that it will make their jobs obsolete.

The trend of ChatGPT is on the rise, with an abundance of cheat sheets, books, prompts, and code examples on how to use it for testing emerging. However, human testers possess a unique advantage that machines cannot replicate.

In this blog, we will explore the unique role for human testers in testing and the various skills that will help them to stay relevant and ahead of the game in the coming years.

 Debunking the myth: Will AI really replace humans?

 

The idea that AI will take away jobs is not new. In the 1960s, many people feared that calculators would take away the jobs of accountants, mathematicians, statisticians, economists, and more. However, if you think about it today, you will easily laugh it off as a nothing more than a joke. Instead of taking away jobs, calculators and other forms of technology have improved all these professions and allowed them to work more efficiently and effectively.

Similarly, ChatGPT can be used to supplement us with a lot of regular day-to-day testing and work-related tasks such as email drafting, test data generation, code optimization, interpretation of code for review, provide comments, and much more.

I was telling the above points to a testing fresher a little while ago and I got this response:

‘Okay, I get it! But ChatGPT can also write code as well as automation scripts. Will it not take away at least some percentage of test scripting or coding jobs?’

This might be a question on the mind of a lot of testers, especially if they are just starting out in the software industry. Once you spend time in the industry, you will soon realize that programming and test scripting related jobs are much more than just code or script generation. They involve a wide range of skills, including:

  1. Thinking
  2. Learning
  3. Social interaction
  4. Investigating
  5. Exploring
  6. Debiasing
  7. Debugging
  8. Analysis, and much more.

These skills can only be assisted by AI tools, but they can never be fully replaced by any AI, no matter how advanced it becomes. Even though ChatGPT can be trained on vast amounts of data and generate test scripts that are becoming impossible to differentiate from good test code. It is not capable of the same level of critical and creative thinking that a human possesses.

Moreover, the human element of interpretation of context, nuances, emotions and critical thinking cannot be replaced by AI. These elements are especially important in fields such as programming and testing, where human judgment and interpretation are critical to the success of a project.

The Exclusivity of Human Testers: What makes them indispensable?

While the list of things that are exclusive to human testers can be long, here are some of the things that are very exclusive to us:

1. Exploring Requirements: Professional testers are responsible for exploring requirements in detail.

They can:
  1. Take notes
  2. Ask questions
  3. Offer ideas and suggestions
  4. Perform investigation
  5. Look beyond requirements
  6. Compare multiple sources
  7. Ensure that the software is meeting the needs of the end-users

While ChatGPT can assist with generating some sample test cases based only on the written requirements. It cannot replicate the depth of analysis and insights that a human tester can bring to this entire process.

Pro Tip: To store such valuable and diverse information, you need powerful tools such as TestGear which provides a powerful test library that helps you store your test documentation and product info in a centralized manner.

2. Bug Advocacy: Professional testers are experts in bug advocacy. They can:
  1. Identify and report genuine bugs
  2. Distinguish between symptoms and root causes
  3. Assess the risks associated with those bugs
  4. Identify use cases associated with those bugs
  5. Investigate the bug by going deep by one or more layers
  6. Communicate the impact of those bugs to stakeholders

While ChatGPT can help in writing generic bug reports based on a symptom, it cannot replicate the nuanced understanding, investigation, and communication skills of a human tester.

3. Usability Evaluation: Professional testers are often responsible for evaluating the usability of the software. While functional correctness is important, it is not the only measure of quality. Usability, accessibility, and other factors are also critical to the success of the software.

These factors include aspects that human testers can evaluate in a unique way such as:

  1. Learnability
  2. Efficiency
  3. Satisfaction
  4. Error Proofing
  5. Utility
  6. Memorability

You can read more about such factors in depth here: Usability Heuristic Checklist - Testing Titbits | Rahul Parwal

4. Question assumptions: Professional testers have can question assumptions and investigate further. They can think critically about the software and its use cases, and able to dig deep to uncover potential issues or concerns. They can also ask questions proactively to find more information about:
  1. Testing mission
  2. User personas
  3. Risks
  4. Market factors
  5. Competitors, etc.

Whereas ChatGPT or similar AI systems are limited to the data that it has been trained on, and it cannot go beyond that. AI systems mostly work on explicit or documented knowledge and does not account for the unspoken knowledge of our social world.

5. Learn from Serendipity: Professional testers can recognize unexpected issues or patterns and are able to adjust their testing approach accordingly. Testers possess the remarkable ability to adapt their approach to software testing based on a variety of factors such as:

  1. Scope
  2. Judgment
  3. Risks
  4. Blocking areas
  5. Dependencies
  6. Available options
  7. Resources
  8. Time available, etc.

Not only do they understand the importance of following documented procedures, but they also possess the practical knowledge of how to apply these steps effectively in different contexts. To adapt all these, you often need flexible tools such as TestGear.

TestGear is a flexible test management tool which works seamlessly with modern bug trackers, task trackers, CI/CD systems and webhook tools. The best way to determine whether a product is right for you is to try it out first.

You can try TestGear for 14 days for free.

On the other hand, ChatGPT is limited to the data it has been trained on, and it cannot recognize unexpected issues or patterns.

6. Systematic Exploration: Professional testers are great at exploring the application beyond any scripted logic. They think creatively about the software and its use cases and are able to find issues that may not be apparent from a purely scripted approach.

ChatGPT is limited to the data it has been trained on and cannot explore beyond that.

7. Testing to Context: Professional testers can tune their testing to context. They adjust their testing approach based on the needs of the:

a. Project
b. Product
c. Team
d. Business

ChatGPT is limited to the data it has been trained on, and it cannot adjust its approach based on the context.

Conclusion

While ChatGPT is a revolutionary technology that can be used to augment certain aspects of testing, it cannot replace professional testers. Human testers bring a depth of analysis, insight, and creativity that cannot be replicated by AI, and they are critical to the success of any software development project. As AI continues to evolve, it is important to remember the unique skills, resourcefulness, and abilities that human testers bring to the table.

To summarize, here is a quick summary of this article in form of a mindmap: https://www.xmind.net/m/mq5aqw

Happy Testing!

Written by Rahul Parwal, member of the TestGear community.