No one told AI to slow down | Issue 8

Welcome to Issue 8 of unfolding ai, we are launching our ‘recommend our newsletter to everyone scheme’. Perhaps not the best marketing title, I should perhaps pop it into chatGPT.

The details are towards the end of the newsletter.

The AI summer lull, it was all over, AI was done. Turns out that no one told that to AI! Absolutely loads of new releases, and updates to just about every section of the AI landscape has been building away over the summer.

Q4 and conference / announcement season is almost here expect it to be a busy autumn for announcements, starting with Apple’s iphone launch as a kick off.

Best regards,

Paul, Co-founder (and newsletter editor)

The key to building rapid success

The most powerful tool for chatGPT (and others) is the copy / paste functionality.

If you are relying on the ‘history’ function to find your amazing prompts, when you get to a few dozen, that’s about when I discovered, there is no search function. It also does not recall if for that prompt you had used any ‘custom instructions’ set in your preferences.

I use notion.so (a smart notebook app) to build and categorise my prompts. You should develop a solution that works for you. This is your growing personal productivity playbook.

As you start to implement these tools into your team, or enterprise. Think how they will be shared, and catalogued. I would suggest you have a session with everyone on the team, to agree

1. Format of how they are recorded

2. Where they are recorded,

3. How when they are used, people can review and rate them

4. Any risks / issues / limitations that people should consider

In the event of everything else, a shared document is a really good place to start.

The impact on the working week

The latest data of applied to a knowledge worker (white collar, business professional) works out at 16 hours of their week would be suitable to be replaced by automation & available AI technology.

Incase you are wondering, that includes C-suite.

The Apple keynote.

Not a lot of information from apple around AI In the latest product announcements A couple of mentions of the use of machine learning sitting behind key applications such as health, Siri and others

Of a little more interest is the upgrade of the internal Apple silicon to include ‘neural’ processor in the watch, and they increased the power of them in the phone. This paves the way for execution of machine learning (and some AI) on local device, required for off network activities.

Siri is definitely looking lackluster at the moment, combined with apple shortcuts there could be a real opportunity for something innovative here from apple, these shortcuts I feel are an under exploited technology.

The Apple segment on ESG was however a masterclass in how to put this into the forefront of your business, and consumers.

Launching our Referral Programme

We are a growing business, and newsletter. We could use your help. We are building out a referral programme to reward you when you do. Rewards will include fun things, like digital artworks, , stickers, discount codes for events that we are hosting, extra playbooks on key topics, and also 1-2-1 Sessions with Unfold AI co-founders.

We will provide excellent value for you as a thank you.

I want to be an AI Prompt Influencer…

We try to always be positive at unfold:ai, one of our guiding principles is to give the best, relevant, and tested advice based on our combined 750,000 hours of industry experience.

We are seeing a lot of ‘ultimate game changing prompts’ being posted to social media.

Here you go, now you can generate your own!

generate me 10 'game changing' prompts for chat gpt that would supercharge my business, format them into a listicle for linkedin. Make them very ‘clickable’ for engagement. 

Provide the actual prompts use square brackets, to denote where I should add custom information.

When we provide a prompt we have the following rules

1. will it genuinely work, and we test them

2. would we use it in our business

3. will it provide gainful repeatable productivity for you, or your business.

One of the getting started pieces of advice we provide is ‘get good help from people who know industry’. If last year your advisor was providing information on metaverse, cryptocurrency, or was an intern. Common sense should be applied.

For AI to be impactful in your business, you need success, sorry to have to provide a warning for you, be safe out there!

AI in Health Care.

Health care is a very complex area for AI. It falls into four broad uses

Case 1 - Scientific

There is a long history of AI and machine learning being applied to a number of ‘at scale’ problems, protein folding, genome mapping, and accelerating vaccine variants and production testing. There are regular releases and information about the application of AI in this space. It has the interesting side benefit, that this rigour, scientific method trickles down to testing, and implementation tools to improve AI robustness, speed and safety.

Case 2 - Professional Diagnosis support

In a much earlier stage is the use of AI as a diagnostic support tool, the most famous of which is probably computer vision use for radiography. Other emerging platforms to support and help with diagnosis are under slow release, test and development. As you would expect this is a careful process, and is also very subject to legislation and practice control.

Case 3 - Workflow

The practice of healthcare is over burdened with manual, and duplicate processes. As much as 60% of a clinical process is time spent on essentially documentation, forms and workflow. AI would be a great assist in this space. Amazon and others are making a play into this platform area

Case 4 - Internet self chatGPT diagnosis

You can google how to catch, solve, and diagnose every disease on the planet, and you can also get chatGPT to do the same. There have been a couple of ‘positive news stories’, where the use of chatGPT has linked together symptoms, leading to good outcomes (by asking better questions of the medical specialists). It’s just a matter of time before a horror story occurs.

I would not recommend this as a use of AI in personal medical practice.

Hobby medicine falls into very high risk of being wrong.

Legal, compliance and regulation news.

Microsoft have joined Adobe in offering indemnity against prosecution from third parties over the use of copyrighted material used in AI training. When using their co-pilot range of AI products, or enterprise services.

This is a reassuring statement. But considering there are no legal precedents, processes or agreements that you can actually be applied regarding AI used in training at the moment, cynically you could ask what are they actually taking as a risk or liability? Or is it just a ploy to sway you from an open source solution ?

Always look behind the headlines for legislation, legal statements from Big Tech companies.

Quick things to check out

If you haven’t completed our two question survey, please do we will share the results next week.

Deeper reading links on AI in healthcare

Did you miss these articles on our blog?

Inside premium this week

• Why data is important to AI

• Think about how to phrase your AI implementation

• What is Expertise Displacement

• The Impact of AI on ESG projects

Plus the usual prompt fun for MidJourney, and dadGPT (tested ofcourse on my own kids)

Subscribe to Premium to read the rest.