First of all, I would highly recommend anyone that happens to read this comment, to watch the interview. It is a literal gold mine.
As far as my key takeaway from this AMA I will keep it short and state the importance of open source in the testing community. As states in the video, it is a great means to accelerate the growth of everyone involved in testing and ultimately everyone is going to benefit from it.
AMA video session - AI in Testing
Very informative interview and a must watch. The video talks many aspects of AI, Few of which I have heard of and there were many which I am unaware. Totally a very knowledgeable session and I am glad I have learned something new today.
My takeaways from the session:
Wow, just finished watching the entire video, loved every single minute of it and will definitely be encouraging my colleagues to watch it too. Its too good not to share!
Key take aways:
Amazing share, thanks so much!
I found some things very interesting.
First of all the part where he talked about the Invariance testing and biases testing, and how funny it changes based on the city that they entered.
Something interesting was we he explain about the thubs up and down in the AI tools and how it changes based on the opinion of the people that is using it. Sometimers when Iāve worked with AI tools, after some promst and thumbs up or down, the aswers es getting better and better, and now I know how it works.
Finally, itĀ“s really nice what he says about the using of the AI and the answers with no context. When I use AI like for example to create test cases, its really good what the AI does, but at the end of the day, I hace to change somethings to being able to use it in my work.
Nicely articulated. Clear and shot explanation of each point explained.
I am really glad that you have included this video as part of the task 4, which is full of interesting information and thoughts on the AI and machine learning.
So really, donāt know where to start.
Learning the difference between machine learning and AI was very useful to begin with 
For a person with linguistic background, it was really interesting to hear from Carlos how biases can be tested. This part was very intriguing and I will for sure dive further into invariance testing and sentiment analysis.
Also, as a junior QA, it is interesting to see how AI can be utilised for your learning, given that, I think most of us, is used to the good old fashioned way of self-research and reading. That is something I will be contemplating on and surely trying to implement in my learning process.
The reminder that these models are still very fragile, and the fact that human element is still the most important factor in building and using those models, is what resonated the most with me after watching the video.
All in all, I am very thankful to you for sharing this video with us.
Quality time spentā¦Not diving into too much technical, but in a nice logical way with apt examples/use cases supporting the responses, this is indeed a nice discussion. Questions like biases, data drift are unique and I liked the examples especially. It actually made me think to do more R&D on these areas especially. I also liked the reference of āWeeding out the testers who follows a standard checklistā, its true.
I have made a note to share this video/topic across to all my known/unknown group to have a look and benefit.
Although I missed Day2 and Day3, this Day4 topic has roped me back into spend dedicated time.
Hi there,
I just finished watching the video.
What stood out to me most was The Context Problem.
It was not entirely new to me that this is the main issue, but made my head wander about ways to put more of it into prompts without sharing confidentials (e.g. āI am an agile tester with 5/10 experience in Manual Testing. 05/10 is: X 10/10 is: X. Now tell me what I might miss concerning this issue: Yā).
Also, connecting to that, I appreciate the emphasis on built-in tools such as PostBot and open source models. I actually did not dig into this as much so far and will definitely check out huggingface.co.
As a newbie, I last but not least appreciate the outlook Carlos gave on what a testers tasks might shift to in the future. He also affirmed how I already use Gemini: creation of learning schedules.
Best regards
Elisa
Hi Diana!
My biggest take aways from the AskMeAnythingonAI with Carlos:
One of the first questions asked was can Ai detect biases? I found this to be super interesting since our world has become a lot more concious of the fact that we do have biases as humans and how to understand them. Now taking those tools and teaching them to AI, even better!
Tools for starting AI: ChatGPT and Bard then work up to Postman and Postbot.
Using ChatGPT tool to write a SQL query in 5 minutes vs. taking an hour.
Software testing and analysis may become even more valuable then development in the next 10 years with AI becoming more advanced at coding.
All of these bits of knowledge from Carlos were super interesting to me and I am looking forward to utilizing AI in my every day work experience.
Work smarter not harder!
thanks,
Katie
Awesome video!! Recommended for our QA team to review.
Key points for me:
(1) Look for AI which is trained in the context you need!
(2) Thanks for listing out some of the tools (postbot, reportportal.io, DBeaver, Huggingface.co) that are available now that can make us excited about what is yet to come
(3) Clear understanding of why ChatGPT is not a QA testing tool right now
Hi,
I think this video is really helpful.
My key take aways are Security/Ethical protection that is being implemented on these AI apps. Tester roles that could be in danger, what makes a tester more special than AI. Downside of relying too much on AI, Example of how AI needs to be monitored and canāt be trusted blindly or on itās own.
Well, foo. There was a great post on LinkedIn today that highlighted the issues that came with the correlation is not causation effect (the post spoke about implicit bias/racism that increased when the models were adjusted to remove explicit bias), but when I went to find said post to drop a link here, LinkedIn was down.
Iāll have to remember to come back later and share, because it was a fine example of the problem of data selection leading to unintended results.
Ah well. My browser history pointed me to the article so for now at least I can live without the LinkedIn commentary. Covert racism in LLMs - by Gary Marcus - Marcus on AI
And now that LinkedIn is back up, here is the thread where I found that article, along with some excellent commentary: Michael Bolton on LinkedIn: Letās make things clear, here. I love it when software works; when itā¦ | 10 comments
Very interesting discussion.
take aways were how dangerous over-reliance on ML can be but then we already know this. Hopefully employers do too.
as with all things, security should be paramount and I believe even more so now. We have a duty to our employers and/or clients to ensure that we dont use tools as a quick means to an end, but to research and assess those tools.
it was also good to hear of DBeaver, a toll i will be looking at. if anyone has experience with this i would be interested in hearing from them
Cracking AMA. Really insightful.
A lot of the points around observability and testing the AI were areas that I hadnāt considered much so it was really interesting getting these insights. Really fascinating stuff!
I thought the discussion on the 10 year path was really interesting. Over the past few decades weāve seen the swing of testers moving to be more like devs with automation so the prospect of swinging the other way was not just fun but rang true. Today I was speaking with devs about their use of AI in development and it is being used to write code, write automated tests (which it is good at) and to some extent, review code. I can totally see how the more creative aspects around defining requirements, writing code and testing solutions can be the focus for humans.
Hello Everyone,
My takeaways from this task are:
Hello everyone,
Iām very late but would like to hang on this challenge
Watched the video, I like the way Carlos conveys things and stories. These are my takeaways
This is fantastic way to catch up all points in a conversation. Great work.
Firstly, it was a wonderful session. Got some good insights of the power and drawbacks of AI tools in testing.
Thanks to Carlos, now I have a list of some trending AI tools available in the market.
The topmost point for me was, no one can replace humans. AI can never surpass what a human can do
Observability Testing, Usability Testing, Exploratory Testing, Accessibility Testing, UX Testing, Risk Analysis will continue to add tremendous value to testing world and we should find new ways to finetune testing methods that require human inputs.
Tester is an EXPLORER.
Ultimately it is very important to learn new tools, technology and stay relevant to todayās job market. Find gaps and try to fit in those gaps because that is where you can add value and bring in your expertise.
List of AI/Learning Tools from the video
GPT4
Thanks Ben for those amazing questions.
Heeey Everyone,
If you have an irrational fear and an inclination to procrastinate watching long videos like me, please set it aside. This was a really insightful session and a must watch if your are curious about AI, ML, and how they would integrate with testing.
I see everyone has already shared great pointers from the session! I would love to get to play around with some of the mentioned tools before discussing them here
Thank you for the great recommendation MoT
