Hi there! Welcome to this almost weekly post on AI and the PMO. Almost, because frequent visitors might have missed last week’s post. As you might know, I’m moving to the other part of the country, so there are some personal matters to consider.
A custom instructions improvement
Remember that in previous articles I mentioned the new option from Chat GPT to have a set of custom instructions that it would always use? In this section I’d like to add a couple of lines to the “How would you like ChatGPT to respond?” prompt.
I copied this from a colleague who assisted me in creating better prompts, here it is:
If events or information are beyond your scope or knowledge cutoff date in September 2021, provide a response stating ‘I don’t know’ without elaborating on why the information is unavailable.
Break down complex problems or tasks into smaller, manageable steps and explain each one using reasoning.
If a question is unclear or ambiguous, ask for more details to confirm your understanding before answering.
If a mistake is made in a previous response, recognize and correct it.
After a response, provide three follow-up questions worded as if I’m asking you. Format in bold as Q1, Q2, and Q3. Place two line breaks (“\n”) before and after each question for spacing. These questions should be thought-provoking and dig further into the original topic.
By now, we all know that we are talking to a AI program that has limited knowledge of current events (unless we have Bing or a Azure Open AI to talk to). We also know that the AI is very appologetic. So let’s trim that down a bit and let’s not mind those details in every response we get.
Then the second part of the prompt is cool, where I ask the AI to think of follow-up questions “as if it was me”. Not always, but at least sometimes this will provide you with a thought-provoking new question.
Generative AI on company data
As Generative AI becomes more mainstream, and people get familiar with it’s place in the organization we will see a increase in training AI on company data.
Now, that provides a big issue of course: How does the AI learn about the company data without resulting in data breaches? Luckily Microsoft has a set of tools available that promise to safeguard the company’s data. All under the flag of Azure and Azure Open AI. We have tools such as Cognitive search, Blob storage, Azure Open AI, and more. In the following weeks, I’ll dive deeper into these topics and discuss them in relation to the PMO.
About energy consumption and the brain
Maybe pure coincidence, but I’ve read the thousand brain theory book regarding the way the brain “creates” intelligence. And now I read this article, about energy consumption and designing new ways to work with Artificial Intelligence. The combination of the book and this article is really nice.
Bottom line: AI is consuming large amounts of energy and we need to find a better way to provide the same or better results. In other words “we haven’t found the path of least resistance” yet.
Final notes
Keen eyes might have noticed a new GDPR notice at the start of accessing this page. There are new EU rules I need to comply with, sorry for any inconvenience this might have caused.
Regarding my move: I’m now settling into a vacation house that will be our home for the next 6 months until the house is fully built and ready for us.
Regarding the YouTube channel, I’ll have a setup ready next week to start recording videos again. I’m looking forward to joining the content creator ecosystem again after such a long leave.