A few AI thoughts on A1
A bit of background
For many years, my role has been to manage Database systems.
Of course, managing Database Systems is far more than making sure that the Data is accessible. It covers security and governance of the data – who uses the data, where the data comes from, the accuracy of the data and for what purpose is it to be used. Tuning the data so that it uses resources more efficiently and returns results quickly.
The only constant is change
I’ve changed Role and Company recently. I am now working at Rackspace Sovereign. We are in a very interesting area, part of the global giant that is Rackspace (so much more than just hosting), specialising in managing systems for companies for whom the sovereignty of their data is crucial.
Rackspace is positioned as an “AI first” company, as demonstrated with a number of high-profile partnerships to leverage our talents as giants in hosting and support with giants in the AI world.
Training and understanding is key
I am seeing an impressive approach towards training, with everyone in the organisation encouraged to train and certify in the use of AI.
One of the major features of the training is just how much emphasis is put on appropriate use, governance and data security.
Understanding and Responsibility
If you use AI, you should work to Understand what it tells you, and take Responsibility if you use the results.
When starting out with AI, the best questions to ask are ones to which you know the answers. That might sound counter-intuitive, but it’s really useful to get a grounding on how accurate the results are.
As an example, I’ve got back into running recently, mainly focussed on getting my 5K time down, as well as considering entering a 10K race. I asked Gemini (Google) what sort of time that I’d likely achieve for the 10K based on my 5K times. There are running calculators that have been around for years that’d answer this question, but it’s a starting point.
Once I had the answer, it seemed accurate, but I wasn’t happy – I had a finish time in mind, so I asked it what time that I needed to be running 5K for the 10K that I wanted. Gemini accurately brought me down to earth with the result and offered to write me a plan.
Now, this would have been difficult using traditional methods, as the race is in just a few weeks. Gemini wrote me a short plan, which exactly matched my expectations – and included the four different workout types that I had already started performing. Well done! A+
Hallucinations
That was a good example, but AI is well known for it’s “Hallucinations” – defined by Wikipedia as
“An AI hallucination is a phenomenon where a generative artificial intelligence (AI) model—such as a large language model (LLM) or image generator—produces confident, plausible-sounding, but completely false, inaccurate, or illogical information.
Aspirational Thoughts
I’m sure that I’m not the only person who researches potential future employers and people. It’s good to know as much as possible about the business and people to see if you would fit in there.
After doing a few searches, I asked Gemini where I worked. It told me that I already worked at the company that I researched. A lovely aspirational thought, but wasn’t true at the time of asking.
It almost seemed like the old “Get someone to say shop five times quickly then ask what do you do at a green light” trick. Most people would say “Stop”.
Office Hours
My favourite hallucination though came up during research for this article. Some years ago (circa 2018), Brent Ozar used to run a regular “Office Hours” video series on YouTube with him and his staff answering user submitted questions. I noticed that quite a few questions got the same response and following a chat with Brent at SQLBits conference, an idea was formed. I duly recorded and published it.
I asked Gemini “Who were the guest presenters on Office Hours” and got a list of distinguished and recognised professionals, such as Pinal Dave and Edwin Sarmiento.
My next question was “How about on April 1st” and the response amazed me:
Luckily, I saved a copy of that response as Gemini now gives a different view. Check it for yourself, I feel flattered by what it reported when I was writing this Post.
Anyway, here’s a link to the clip, make up your own mind.
Checking the work of AI
It’s not exactly “Billy Liar” or “Jay from The Inbetweeners”, but once again, you need to do your own reseach and mark “your” own homework. If you use the results from an AI enquiry then you are responsible for those results.
Careers in AI
There is a lot of concern about job losses as a result of AI. It seems that major corporations are using the line of “AI has made the organisation more efficient” when laying off staff.
The line generally used to explain the benefits of automation is probably more accurate right now – there are parts in everyone’s work that can be assisted by AI which provides more scope to do more things.
As an example, I had to write some code to reflect a change in file location recently. I could have written a script to perform this work myself, but I asked for AI assistance. The generated code was very good and it was prepared in far less time than it would have taken me to write it.
I’m talking about T-SQL Code, something that I understand. Now, at my previous role, I was partly responsible for performing code reviews – so I reviewed the AI generated code before running it, and it was Approved.
Quite different from the “Vibe Coding” that if oft bragged about. Marvellous stuff if you know what it’s doing and can support it – until it goes wrong, needs to be debugged, checked for security flaws. Makes me think of “Jive talkin’ you’re telling me lies” by the Bee Gees (or George Michael, take your pick). Actually, most of the lyrics are oddly appropriate here.
https://genius.com/Bee-gees-jive-talkin-lyrics
Careers
Going back to careers – there are lots of AI related roles out there. I’m not just talking about the £100,000 Prompt Engineers.
My own role of Database Administrator, keeping Databases fed and watered and tuned like fine racehorses is to be extended to looking after AI systems. They have the same basic needs including Security, Management and Tuning.
I have people in my Network who are involved in visualising the role of AI in their organisation, and also those who work to encourage and enforce AI Literacy and Safety.
These are not brand new Roles or specialisms, these people have come from being highly respected in their previous roles in producing Educational content and Editorial literacy.
Podcast Recommendation
Finally, I’d like to recommend a Podcast that I’ve really enjoyed listening to. It’s called “Married To The Machine” and it’s described as “Everything you ever wanted to know about artifical intelligence delivered in a non-techy way”. It’s a really good listen, and great to get up to speed with what’s happening out there. Probably the only Podcast that I’ve “binged listened” since “Sandra” (a thoughtful little tale about a woman who goes to work in a call centre, answering all those questions that people asked their Alexa devices).
Thanks for reading!
Nigel.
