There was an initial wave of AI frenzy a few years ago, with every tech product producer, tech company, tech commentator mentioning AI as much and as often as possible, it seemed a passing fad, where they’d just substituted the easier to understand word AI for algorithm, as that’s what AI seemed to be at that point.

However, now ChatGPT and other implementations of large language models are seemingly everywhere, things are changing. Even though the contents of ChatGPT et al are mostly ‘scraped’ from existing content they’ve somehow managed to get a very user-friendly and fairly easy to use front end onto these things. User interaction with them is quite a simple process, so millions of people have already been trying this out, either in ChatGPT, in Bing or in other places. They have a conversational front-end which makes them easy to use. Open AI released ChatGPT apps in May 2023.

As when most new major changes come, it will take a while to see how this all pans out. It seems a bit similar to when the ‘Net suddenly went from something geeks (e.g. IT people) used to suddenly everyone was using them. This was partly when broadband allowed easier connection (rather than the old dial up route) and really took off when everyone starting having a mobile phone – iPhone, soon followed by Android phones.

AI includes things such as Machine Learning (used in recommendation systems by Amazon and many others), Large Language Models (used by ChatGPT et al), advanced Speech Recognition (such as Alexa and Siri), self-driving cars and generative tools such as ChatGPT and DALL-E.

The way ChatGPT and similar tools get their content is mostly to scrape them from existing content, so it is using people’s content, which opens up copyright issues and possible legal action from copyright holders and content creators. Reddit want to start charging big companies for scraping its vast content.

A major worry for the world’s workers is what impact AI will have on their jobs and careers. You would hope that companies would let the dust settle for a while to see how AI performs now that we have these tools such as ChatGPT. However, in the UK, BT have already announced plans to get rid of 55,000 jobs with a fifth replaced by AI. It seems like they’ve jumped the gun to get a seemingly cutting-edge mention of AI in their announcement – it will be interesting to see how that all pans out.

Another concern with AI is the vast amount of resources it takes to populate its models and the data on which it is based. I would imagine its on a similar vast size as Google Search, so there must be many eco concerns.

Another consideration is that the public in general tend to have the mindset that whatever the computer says must be correct, whereas people who work in IT know that its best to take what the computer spits out with a pinch of salt. We can already point to incidents where AI has not got it quite right, such as this story on the BBC website about a weapons scanner and this set of tales of woe from Jumpstart.

As a software tester, I would hope that AI will allow me to use various AI-assisted tools to improve the efficiency of my work, possibly partly on the admin side and maybe also with some testing tools. It would be handy to have tools that help with creating test scenarios, images for testing with, user scenarios, user journeys, test data etc. Therefore, AI would become a tool in my testing armoury, rather than replacing me as a tester – at least that’s the plan anyway! As someone already mentioned to me, try sending AI into a meeting or get it to figure out what’s what in a Figma design. Talking of AI tools, I’ve already used the ‘Bing Image Creator’ to create some interesting images, which I might one day use in some tests, such as the image for this post – which is generated from the text ‘A frog in a blue suit and tie’.

Earlier in 2023, Microsoft upped the ante on AI by more or less stating that their Bing tool would outdo Google’s search as it had an alliance with OpenAI and was using ChatGPT to power Bing search tools. Google fired back with their Bard tool, which didn’t get off to the best start. Since then, Google has announced Generative Search, where you can search as before or ask it something and it should perform a chat type answer.

And finally, comes the question, was this post written by AI? Erm, no… but I might get it to help me with writing the next one.

Some thoughts on what I’d like AI tools to enable me to do:

– As a tester – create a whole batch of test data based on some criteria I supply

– For listening to music – create a 5 hour mix based on this track OR create a 5 hour extended mix of this track

– For creating music – create a set of 909 house drum patterns at 120 bpm

Interesting links

Microsoft will ID its AI art with hidden watermark

Pentagon attack hoax triggers brief market sell-off

Workplace AI: How AI will transform the workday

Bill Gates says AI winner will be whoever creates a personal assistant