Welcome!

It’s getting hard to stay on top of relevant AI news, and to cut through all the noise. I get about 30 AI newsletters a week, and they all repeat each other.

I am trying my best to keep this newsletter different, and unique to planners in New Zealand, so if I skip a week, it’s because there isn’t as much going on.

This week’s newsletter is focused on the usual interest pieces, plus a bit more:

• 👨‍🏫 ChatGPT Prompt of the Week: Unravelling Planning Policy

• 🔍 An AI District Plan for Kaipara?

• 🚀 Hotjar AI: A New Survey Sidekick

• 🌟 Learn About Generative AI with FranklyAI Live

• 🌐 From Around the Web: Things to Know

• 🍗 AI Roast of the Week - Public Hearings

Also - a quick poll:

👨‍🏫 ChatGPT Prompt of the Week: Unravelling Planning Policy

We'll provide you with an example prompt each week that you can use with ChatGPT.

This week's prompt is focused on dissecting and interpreting the objectives, policies, or criteria within district or regional plans. Plan provisions are sometimes formulated with complex language that may be difficult to comprehend for the general public.

The prompt we're sharing aims to address this challenge. It guides ChatGPT to read through each objective, policy, or criterion, compile them into a table, and provide explanations in understandable language. It further helps in identifying potential ambiguities and points of misinterpretation, and how these could be exploited. Additionally, the prompt assists in synthesising the information into a concise and coherent introduction to the zone (if relevant - the third step is optional), articulating its intended purposes and goals.

If the explanation doesn’t sound right, perhaps it’s back to the drafting board! You could even look to run the prompt in reverse, and give it a written description of a zone, and ask it to formulate provisions in reverse.

This prompt is invaluable for urban planners, policy analysts, development professionals, and any stakeholders who need a clearer understanding of plan provisions and what they mean.

Example prompt:

Below are the [objectives/policies/criteria] for the [zone name] in [location name], under the [statutory document name].

[Insert provisions here]

I want you to:

(1) Read through each [objective/policy/criteria], add to a table, and explain within the table in layperson terms what that [objective/policy/criteria] is setting out to do.

(2) In a third column within the table, explain how the [objective/policy/criteria] could be misconstrued or misinterpreted, with alternative interpretations for the [objective/policy/criteria]. Explain how someone nefarious would try to find a hole in the [objective/policy/criteria] to achieve what should be unsupported development (based on a plain and ordinary meaning of the provision).

(3) Write a model introduction to the zone, using the [objectives/policies/criteria] for context. Explain what the zone is setting out to achieve. Pretend you're describing the perfect zone that has met all of these [objectives/policies/criteria].

I ran the prompt using the objectives for the Coastal- Marina Zone in Auckland as an example:

Here is an example output from this prompt:

The zone description was way more colourful than I expected, and to be honest, a touch better than the drab descriptions we often see in district plans:

I was curious whether I could then go on and provide ChatGPT with the policies for the zone, and ask whether there were any hierarchy issues, i.e., did any of the policies appear to not integrate well with the objectives:

The results were great (well done to whomever wrote this chapter of the Auckland Unitary Plan!)

Got to love that last line - who does Chat think they are, a planner!?

Remember, while AI can be a useful tool, it's always a good idea to cross-check its advice and continue to use your professional judgement.

We hope you find this week's prompt useful, and as always, we'd love to hear about your experiences in using it!

🔍 An AI District Plan for Kaipara?

Nick Williamson from Fluid Industries has embarked on an interesting venture to integrate AI Large Language Models (LLMs) into resource management planning.

He is working to draft an entire District Plan for Kaipara within a span of just 28 days, utilising GPT-4's capabilities, using a bit of Bing and ChatGPT.

I see this to be a daring test of just how effectively AI can streamline and expedite policy development, a process well known for its complexity and lengthy timelines.

This isn’t just about speed, though. Nick argues that integrating AI into the drafting process could enhance the quality of the District Plan. Through adept data analysis and an ability to digest a wide range of documents, the AI can potentially offer insights and suggestions that might be overlooked.

Moreover, with the extensive public documents and historical data that often need to be reviewed when creating a District Plan, AI’s ability to quickly process this information could lead to more comprehensive and well-rounded policy development.

You can follow Nick’s progress on his LinkedIn and website. As of today, he’s up to Day 7 of 28.

🚀 Hotjar AI: A New Survey Sidekick

Whether you’re steering a business or undertaking community engagement, feedback is our North Star.

But let’s admit, crafting perfect surveys and analysing heaps of responses can be as tricky as navigating through Auckland during peak hour.

Enter Hotjar AI, the suave assistant we’ve been yearning for! It whizzes up surveys with a click. Just whisper your goal, like “assessing public opinion on new raised pedestrian crossings in St Heliers”, and voilà! Hotjar AI conjures up the right questions for your noble cause.

But wait, the magic doesn't stop there! Delving through open-ended responses usually takes eons, right? Fret not, Hotjar AI’s automated reports dish out summaries, quotes, and actionable next steps, like “maybe just put them in and don’t have a community meeting’.

Now, why is this great? Engaging with the community gets a boost. We can swiftly roll out surveys and make sense of the feedback. Instead of drowning in data, we’ll have insights at our fingertips, ready to mould our cities into vibrant spaces that resonate with the people. Moreover, GDPR and CCPA-compliance mean that data safety is as tight as most organisations need.

You can dash over to Hotjar and start for free. For a subscription, they’ll even let you keep your survey responses forever.

Hotjar AI is more than features – it could help lead to more frequent, empathetic, data-driven planning. However, let’s not lose the human touch. While AI is a nifty tool, coupling it with personal interactions ensures we truly understand the pulse of the communities we serve.

🌟 Learn About Generative AI with FranklyAI Live

Mark your calendars for a riveting new weekly webinar, "FranklyAI Live - Understanding the Power of Generative AI (ChatGPT) in Your Role and Organisation", slated to kick off this Wednesday, 21st June, from 11:00am to 11:30am.

Spearheaded by the enterprising Matt Ensor, CEO of FranklyAI and Chair of the AI Forum’s working group on Large Language Models, this virtual event promises to unravel the mystique behind generative AI.

From a lucid introduction to its inner machinations to practical methods for integrating AT tech within various roles, this session is tailored to boost innovation and efficiency across your organisation.

But that's not all - Matt will also guide attendees through a candid exploration of risks, encompassing aspects like accuracy, confidentiality, and privacy. The dynamic format boasts enlightening presentations, hands-on activities, and ample opportunities for stimulating discussions and Q&A.

Register now and secure your front-row seat.

🌐 From Around the Web: Things to Know

Below are the top things I’ve read online in the last week that I think you should know about in the world of AI.

1. Microsoft's Orca: A Leap Forward in AI Reasoning
   The AI research community is buzzing about Orca, Microsoft’s new 13-billion parameter model designed to emulate the reasoning process of large foundation models. Orca learns from rich signals, including explanation traces and step-by-step thought processes from GPT-4. It significantly surpasses conventional state-of-the-art models in complex zero-shot reasoning benchmarks like Big-Bench Hard (BBH) and demonstrates competitive performance in professional and academic examinations such as the SAT, LSAT, GRE, and GMAT. This advancement in AI reasoning is a promising stride in improving the capabilities and skills of AI models. Read more

2. Meta's Voicebox: A Breakthrough in Generative AI for Speech
   Meta’s Voicebox is a groundbreaking model that generalises speech-generation tasks with state-of-the-art performance. Unlike earlier generative AI for speech which required specific training for each task, Voicebox uses a novel approach called Flow Matching, learning just from raw audio and transcriptions. This versatile model can synthesise speech across six languages, perform noise removal, content editing, style conversion, and diverse sample generation. Voicebox’s potential applications range from bringing speech to those who can't speak to effortlessly editing audio recordings. However, due to the potential risks of misuse, Meta is not making the Voicebox model or code publicly available at this time. They are sharing audio samples and a research paper detailing the approach and results. Read more

🍗 AI Roast of the Week - Public Hearings

Each week, we’re going to include in this newsletter something funny ('roast' section).

We’ll try to keep it light-hearted and focused on general planning-related topics rather than targeting specific individuals or organisations - we’re all friends here!

The End (for now)

Thank you for reading our seventh issue of AI-mpowered. We hope you found the content valuable and inspiring. You can find all issues here.

If you enjoyed this newsletter, please share it with others who might also be interested in what we’re covering here.

We would also love to hear your feedback and suggestions for future newsletter topics.

If you have any specific prompts you'd like to learn about or new tools and resources you would like access to, please don't hesitate to let me know.