- April 10, 2017 at 6:49 am #15935
Though it’s a controversial statement, I will be on AI side since I’ve seen things where AI+Automation did much better than us Test Engineers. When I ask this question I usually get answers
1. AI can perform GUI testing?
2. Can AI perform exploratory testing?
3. Can AI design Test scenarios like we do?
This is the most hilarious question to me
4. Can AI talk to a developer and force them to do as we say?
For questions 1,2,3 my answer would be YES. Machine Learning and Cognitive
Systems evolving so quickly. To automate GUI (visual based testing) once I employed a tool called ‘APPLITOOLS‘ & I was awed looking at the unexpected features (you can try free trail as well). But later I have seen Microsoft cognitive services, IBM Watson I was much impressed. Like I said there are few tools which can design test cases if you give UML/DFD/BPD diagrams as an input. You may call this a Model Based Testing. But the purpose has been served right?
I found few interesting sources on Google
This is just my opinion, I’m not dissing or discouraging anyone or any profession. As a Tech Lover I usually try things which saves time. Let me know your opinions. If I am wrong then I can be correctedApril 13, 2017 at 2:55 pm #15980
There is a webinar on this very topic happening next week with Paul Gerrard presenting:April 19, 2017 at 1:55 pm #16007
My knowledge on the capabilities of AI may be limited, but there are a few critical defects that I have found in the past, which I cannot imagine how AI would have found.April 19, 2017 at 2:17 pm #16008
Critical defects like what Archana? Could you share few scenarios?April 19, 2017 at 4:12 pm #16012
Artificial Intelligence (AI) lacks innovation since it can only follow the rules. Humans, on the other hand, step out of the box and create genuine innovations.April 26, 2017 at 11:58 am #16087
Could you share few scenarios?
Here is one of them.
This particular website I tested gave 10% of amount spent back to the user on providing a review. The review would be available after successful completion of a training session and release of payment.
Upon testing, I realized that the review link was created as soon as the session was added to cart (even before the payment). So, I could keep adding sessions to my cart, and give reviews without paying anything. Thus getting credits for nothing.
The way I was able to do this was – add a session to cart. Open a previous review and change a number in the URL which followed a fixed pattern.
When I describe the issue, it may seem very easy. But I cannot imagine how AI could have possibly find this issue.May 4, 2017 at 10:57 am #16206
Although I am convinced that AI can (as yet) not replace the human mind – remember we have to use the biggest computer available to win a game such as chess or from the best human players – the computer can do nothing else: just play chess – the humans can do lots of other things.
That being said the example from @archana is something AI could easily do. URL tampering is very common and can easily be programmed, we just need httpPost. (Big) data analysis will easily detect the fixed pattern.
Lacking creativity is a big problem for AI though, AI only does how it is programmed – and thus depends on the knowledge and creativity of the programmer that coded it. Thinking out of the box or creatively handling new situations is out of the question for AI.
You must be logged in to reply to this topic.