AI for Business: Tools, Trends, and Strategies – RCP Chambers of Commerce

Studio director Mark Sund shifted gears to give a presentation about general business uses AI. Topics revolved around small companies using AI to improve efficiency and danger points to watch out for.

AI for Business – Tools, Trends, and Strategies

For the RCP Power Lunch / Chambers of Commerce, Cincinnati

18 October 2023

Presenter

Mark Sund, Studio Director, Sund + Co

When I’m not giving presentations about AI, I work for clients to generate leads, acquire customers, and support their marketing and sales teams.

What Changed?

Evaluative AI was the first to arrive: it needs a lot of input to come up with one output: a medical diagnosis, credit evaluation, etc.

Operative AI produces about the same amount of output as it got as input: examples are chess-playing engines, assembly-line robots, and self-driving cars.

Generative AI takes a small input (aka prompt) and generates a large output: an image, video, writing, computer code.

Today’s Topics

  • Use Cases

  • The Tools

  • Prompts

  • Summarizing

  • Non-Public Information

  • It’s a Process

  • Issues: Training, Biases, Copyright, Business Risks

Use Cases

Just about everything you do with generative AI falls into one of these categories:

  1. Brainstorming

  2. Research

  3. Summarizing

  4. Creating

Tools

Words

ChatGPT – great for writing
Perplexity – great for research; also writing
New Bing – great for research
Poe – great for chatbots
Bard, Llama – not really ready for prime time

Graphics

Dall-E – great for brainstorming; same account as ChatGPT
Midjourney – harder to use but more realism
Stable Diffusion – harder to use but more realism
Adobe Photoshop (Generative Fill) – incorporating generative AI into an existing product

Summaries

Summarizing tools are exploding – they can summarize a long file or a long meeting! Here are just a few…
Jasper.ai
Copy.ai
Fireflies.ai
Grammarly – incorporating generative AI into an existing tool

Integrated

Jasper, WriteSonic, Copy.ai – designed to manage word-oriented marketing campaigns
Microsoft Designer, Adobe Express – designed to manage visually-oriented marketing campaigns
Hubspot, Salesforce, etc – incorporating AI
Microsoft Office etc (Copilot) – incorporating AI

Nearly everything is incorporating AI these days!

Tool Examples

Technical difficulties at the venue prevented us from switching to live demonstrations – here described things in ChatGPT, Dall-E, and a little about Microsoft Copilot.

Prompts

Prompts

We didn’t spend much time on prompts, because there are so many examples out there – follow them or experiment and improvise.

Another reason to not spend much time on prompts: there is no perfect prompt, everything comes back a rough draft.

Include in Your Prompt

  • What to write about or design

  • Style / tone / audience

  • Length or use case

  • Background – caution: nonpublic info! See below

  • Words to use or avoid and other optional things

Prompt Examples

You can’t always get what you want from prompts – a little later we talked about learning to live with that – the process – and necessity of humans taking over at some point for editing.

Summarizing

Summarizing

Many text-based AI tools can summarize documents – paste them into the prompt or upload (for tools that offer that feature), and tell it the style and length you need.

Video and audio-based AI tools can summarize meetings – create your persona and schedule it to attend the meeting.

Whether I recommend doing these (due to quality and giving up non-public info) is a different question! Discussed more below.

Risk: Entering Nonpublic Info

The Risk of Entering Nonpublic Info about Your Company or Client

If you enter nonpublic info in prompts or a brand kit, that info is now outside your company or client. The AI will train on it, to use as background for answers it gives to other people. Be sure you know whether the AI will train on your inputs, and whatever else they reserve the right to do with it.

Company Policies Regarding Entering Non-Public Info into AI Tools

Consider your company’s competitive position.

Consider your NDA or confidentiality agreement with the company.

Create a policy for employees and members of the company.

Process

It’s not a prompt, it’s a process

That’s our philosophy

Process – AIs are Like:

An intern or entry-level employee
Giving someone too tight a deadline
A natural liar
An incurable pleaser
A brainstorming partner
Managing a team

Professionals using AI are taking on more of a management role.

One Example – An Ambient Music Channel Cover Image

This image is great – the lighting is amazing, composition and perspective are too. But click and zoom in and tell me what you see.

The good and the bad here are typical of AI – both graphics and the corresponding things in writing.

Training, Biases, Fake Details, Errors, Copyright Issues, and the Need for Editing

Training and Biases

Inside the Secret List of Websites that Make AI Sound Smart

Things that make you go hmmm…

www.washingtonpost.com/technology/interactive/2023/ai-chatbot-learning

Fake Details

The AI Tools Making Images Look Better

Details, even when wrong, increase trust – has long been known in courtrooms, where juries believe witnesses giving testimony with lots of details, even if they’re wrong, over witnesses with less detail, even when it’s correct

www.quantamagazine.org/the-ai-tools-making-images-look-better-20230823

Mis-Categorizing and Other Errors

Described mistakes ChatGPT has made in answering about biological kingdoms (high school level biology stuff!) and non-SQL database.

Also the slide above where we looked at a beautiful image with strange details.

Copyright

Content is eligible for copyright protection to the extent that original, skilled, human work was used to create it.

Described what this means for the content the AI tools train on, and what it means for content we create using AI.

Editors Needed

We’ve always needed to review and edit design and writing by humans. We have biases too, and the brand we’re working for has an angle.

AIs allow us to be more productive, yet require more oversight. So professionals and managers need to keep in mind that the time savings are partly taken back by longer and more careful oversight and editing.

Take-Aways

Process Summary

Play around first
Give it the facts: company, products, services
Tell it style and audience
Prompt, revise, and repeat
Take over for human editing

What Kind of Efficiency Gains Can We Expect?

AI speeds up the early stages, but requires more human management after that.

What If Our Marketing Agency Uses It?

Make sure they know that the NDA they’ve signed applies to what they tell AI.

Big Picture for Businesses

Curate and iterate

Avoid nonpublic info

It’s like managing a team

Wrap-Up

Thank you

To the RCP Power Lunch; the Evendale, Forest Park, Norwood, Reading, Springdale, and Woodlawn Chambers of Commerce; and Kathy Walters :-)

With Input From

Thank you to my colleagues for ongoing discussions of how we-all use AI:

Sarah Crawford, Creative Director, Sund + Co

Dino Pelle, Digital Director, Engage321

Q & A

In addition to discussion during the talk, we had time for a little more.

For more about Sund + Co’s lead generation, customer acquisition, and support for marketing and sales teams, visit us at www.sundco.com


Lead Generation and Customer Acquisition

We generate leads for clients and help them acquire customers.

Find out more


Need to support your in-house teams with sales and marketing materials, a new trade show booth, marketing automation, or ongoing lead generation delivered right to your CRM?

Previous
Previous

Blog Articles and Whitepapers — Seidon

Next
Next

AI for Marketing: Latest Strategies – American Marketing Association