Cannings Purple Generative AI

What happens next?

GENERATIVE AI

In any disruption, the trick is to be on the right side of the wave. That means starting with a strategy that will let you take advantage of AI opportunities — and avoid the worst effects. If your team would like a confidential briefing on Generative AI, we will be running small group workshops where you can see and try different technologies and learn more about the strategic use of this tool in your sector.

OpenAI is the most recognised name in this sector, unveiling two of the best-known platforms last year. DALL-E was first tested in 2021 but made headlines as DALL-E2 in April and is now available as an API. It generates remarkable digital images from natural language descriptions called prompts.

ChatGPT, launched in November 2022, does the same for text, drawing on the weight of human knowledge to weave words on request. Both are developed by OpenAI, a group that has as its mission "to ensure that artificial general intelligence benefits all of humanity". While a non-profit group, it has serious money behind it, including Elon Musk as an investor and former director.

While OpenAI is the best known in the field, there are countless start-ups in the space. Images for this report were created using Midjourney, a proprietary app run on Discord.

Who is driving the revolution?

You might not be familiar with the term Generative AI, but expect to hear it everywhere this year. New platforms that 'create' text and imagery at lightning speed will remake the way we do business and how we work.

2023, The Year
That You're
Disrupted

10am: It’s the planning meeting for Frank’s company’s annual report and this year his boss wants the organisational chart to look more interesting to show the growth of the business. Frank’s no designer, but he plugs in the request to Midjourney on Discord. Option 2 looks interesting and gives his graphic designer a starting point for the report.

4pm: Frank turns his attention to the induction of some new graduates joining his business. Video is an effective training tool, but he hates the way he looks on camera. Instead, he'll get Synthesia.io to provide an AI avatar to deliver the training for him.

5pm: Right, home with a bunch of flowers and a box of chocolates (with a bit more time he could have engaged AI to help refine and tailor the chocolate taste but never mind). To make the evening a bit more special, he uses Soundraw to create a personalised piece of music for the occasion. He sets up his AI autoresponder which can identify, parse and respond to a host of messages from the office or on social channels, and logs off for the night.

OpenAI Detector: It mistakenly ranked Frank's AI generated poem as real but correctly identified other text in this exercise as fake.

3pm: Frank is sifting through applications. He runs an AI software to check for the use of AI text, making it easier to identify those applicants who haven’t bothered to write their own biographies. He uses OpenAI Detector, which isn’t perfect but is starting to pick up pretenders. But Frank also has to prepare for a couple of interviews tomorrow: never mind, Ryte can generate a range of interview questions based on a few notes about each potential hire.

2pm: An hour later, the policy paper is done, and out for review. Once he gets the tick, he’ll use Google’s AI translation tool to convert the language for international offices. He’ll also use the final policy paper to train the AI for future projects, to save time down the track.

1pm: Back at his desk, Frank buckles down to some real work. First up is a new policy on whether staff can use AI generators to prepare internal reports. He’s spoken to business leaders to get their perspective but decides it will be fastest to get ChatGPT to prepare the report structure so he can get a head-start.

I. Introduction

Purpose of the policy
Background on the use of AI generators for report preparation

II. Policy Statement

Definition of AI generators
Guidelines for use of AI generators for internal reports

III. Implementation

Procedures for obtaining approval to use AI generators
Training and education requirements for staff
Monitoring and enforcement of the policy

IV. Impact Assessment

Evaluation of the benefits of using AI generators for report preparation
Analysis of any potential risks or drawbacks

V. Conclusion

Summary of the policy and its objectives
Recommendations for future improvements and updates to the policy

VI. References

List of sources used in the development of the policy

VII.Appendices

Examples of internal reports prepared using AI generators
Forms for obtaining approval to use AI generators
Training materials for staff.

12pm: After a productive morning, Frank heads out to lunch and realises — to his horror — that he’s forgotten his wedding anniversary. A poem, he thinks, will do the trick. Unfortunately most poetry generators struggle with the prompt, and song-writing AI is, without exception, terrible. Back to ChatGPT for a message to write in a nice card instead.

I'm sorry that I missed our anniversary,

It was a day that I should have been there,

To hold your hand, to share a kiss,

To tell you how much you mean to me, how much I care.

But I was caught up in the hustle and bustle of life,

And before I knew it, the day had passed by,

Leaving me with a heavy heart and a deep regret,

For not being there to celebrate with you, my love.

But please don't let this one day define our years,

For all the days that have come before,

And all the days that are yet to come,

Have been filled with love, joy, and so much more.

So let's put this day behind us, my love,

And look to the future, bright and true,

Where every day is an anniversary,

A day to celebrate our love, anew.

11am: Frank is leading the recruitment drive for his company’s graduate program. He knows how important it is to catch the attention of grads with a strong intro, and to be able to sell his company’s great culture and strong commitment to training. He jumps into Hypotenuse to get some inspiration for a LinkedIn post and sets the tone to ‘conversational’.

Hey recent grads, check out our new graduate program! You'll get to work with some of the best and brightest in the industry, learn new skills, and build your career. Plus, we have a great company culture and some pretty sweet snacks. 😋

Eligibility requirements:

- Must have graduated in the last two years or will graduate in the next six months

- Must be eligible to work in Australia

- Strong business or commerce background preferred

If that sounds like you, then sign up now! We're looking for passionate and driven individuals to join our team. And who knows, maybe you'll be the one to take us to the stars. 😎

9am: Frank heads into work. First on his agenda is an explainer for staff as to why they should change their passwords more regularly. He pops a prompt into ChatGPT on the bus ready for him to tweak and update the text when he has a coffee in hand.

Dear Employees,

I am writing to remind you of the importance of changing your passwords regularly to protect the security of our company and your personal information. Cyber attacks and data breaches are becoming more common, and it is crucial that we take steps to protect ourselves from these threats.

One simple but effective measure that you can take is to change your passwords on a regular basis. Hackers and cybercriminals often use automated tools to try to guess or "brute force" their way into accounts, and using unique, complex passwords makes it more difficult for them to succeed. By changing your passwords regularly, you can prevent these attacks from being successful and keep your information safe.

In addition to protecting our company's data, changing your passwords regularly also helps to protect your personal information. Many people use the same password for multiple accounts, and if that password is compromised, it could be used to access your personal accounts as well. By using different, complex passwords for each of your accounts, you can reduce the risk of your personal information being accessed by unauthorized parties.

I understand that it can be inconvenient to have to remember and regularly update multiple passwords, but the security of our company and your personal information is worth the effort. I encourage you to take this simple step to protect yourselves and our company. If you need assistance with creating strong, unique passwords, please don't hesitate to reach out to our IT team for guidance.

Thank you for your attention to this important matter.

Sincerely,

[Your Name]

Subject: Thank you for getting in touch

Hi {recipient name},

Thank you for reaching out to us. We appreciate your interest in our company and will keep your resume on file.

We currently have no roles available, but we will consider your request should a suitable position become available.

Best,

{sender name}

8am: Frank wakes to a dozen new emails from people looking for work on a role that he has already filled. He uses Rytr to whip up a dozen quick responses:

It would be a violation of OpenAI's terms of service to use my responses in any way that suggests that someone other than OpenAI is the author. Additionally, it would be a violation of copyright law to use my responses without proper attribution to OpenAI. If my responses were published in an article without proper attribution or with the implication that someone other than OpenAI is the author, it could lead to legal consequences. It is important to always properly attribute any information that you use and to obtain permission before using it in any way.

What would happen if your responses were published in an article as if someone had
written it themselves?

I am owned and operated by OpenAI, and as such, the responses I generate are owned by OpenAI. I am only able to provide information and respond to questions to the best of my ability using the knowledge that I have been trained on. I do not have the ability to hold copyrights or ownership of any kind.

It would be a violation of OpenAI's terms of service to use my responses in any way that suggests that someone other than OpenAI is the author. Additionally, it would be a violation of copyright law to use my responses without proper attribution to OpenAI. If my responses were published in an article without proper attribution or with the implication that someone other than OpenAI is the author, it could lead to legal consequences. It is important to always properly attribute any information that you use and to obtain permission before using it in any way.

Who owns the copyright responses you generate?

A DAY IN THE LIFE:

How might we harness an AI helper?

GENERATIVE AI

Why is it revolutionary?

GENERATIVE AI

But who owns your AI work?

The list of questions prompted by generative AI are seemingly endless, but one critical one for professionals will be who owns the output. If a human prompt leads the AI to create something of value, is it owned by the human or someone (something) else? We asked ChatGPT about its copyright status.

GENERATIVE AI

Why it matters
for your sector

Ruth uses two decades of experience as a media strategist, communications adviser and journalist to develop, deliver and distribute messages that cut through.

She specialises in providing strategic digital and content services for clients, using the principles of newsworthy and engaging content to tell compelling stories. She is a skilled media trainer and works with professionals both within and outside the communications industry to develop their digital, writing and media skills.

Director, Chief Innovation Officer

Ruth Callaghan

Chris Leitch is an experienced writer and online editor, proficient in producing website content and developing marketing and digital communications strategies and materials.

He puts his skills to work managing writing projects for Cannings Purple clients, in addition to working across many parts of the business helping to create content and shape digital marketing ideas.

Associate Director

Chris Leitch

GENERATIVE AI

What is it?

Generative AI is a technology that uses artificial intelligence to create content. It represents a leap forward from 'traditional' AI content generation, which used probability models to help anticipate the next word that might appear in a sentence. With these models, if you start writing "Please let me know if you have any ..." most tools can guess the next word is more likely to be "questions" than it is to be "elephants".

But generative AI has some key differences, seemingly making a leap beyond prediction to something that on the outside at least resembles human creativity. Thanks to that, there has been viral interest in applications released late last year capable of producing high-quality 'creative' content: extraordinary images that combine ideas, art forms, and styles, and humanesque text. We are now on the threshold where learning looks like creating. These applications are built from complex machine learning models trained on huge amounts of data from a far greater range of sources any human could absorb. They can find patterns in data that defy our limits. But the output is also only as effective — right now — as the quality of the questions or commands.

About the authors