ChatGPT Prompting for Innovation Teams

, ,

ChatGPT Prompt Engineering for Innovation TeamsNow innovation Teams have a new way of using AI and ChatGPT for better innovation results. Dramatically reduce innovation time by 95%, more than quadruple ideation performance, and get a new way of validating your decisions. Innovation task-specific prompts have been developed by the BlueCallom team. Examples are shared here, helping teams to build their own.  To set the expectation: While there are gazillion amazing tips and tricks on how to create better prompts for ChatGPT,  this post is dedicated to INNOVATION MANAGERS & TEAMS. By the way, the old saying “garbage in, garbage out“… still as important as 50 years ago.

Prompting for Newbies

If you are a savvy “prompter,” you can skip this paragraph. For those who are not that familiar with how to enter good questions into ChatGPT, this might be helpful.  ChatGPT is the most advanced way to search, and get clear answers to virtually any imaginable question that you could ask “The Internet”.  You can obviously ask a question by typing, “Who are the biggest competitors of Google? You get a decent answer. But when you are more specific, asking for competitors in the AI space, you get a different answer.  And if you go further you could ask:
“I want you to be my market research assistant.
I will tell you about my research needs and you will create a table with the following columns:
Ranking | Company name | company location | main business focus.
Sort the list by Company Name in alphabetical order
My research needs are as follows:
Find the 10 most relevant AI companies that are competing with Google.

You will get a nicely formatted list.
Now ask a very simple follow on question:

“Show me the same list for European competitors.”

You notice it kept the dialog between you and the AI and completes the task with all the previously defined features.

The ability to write those specific requests like the first one and know how to possibly follow up is called “prompt-Engineering’ The prompt is a well defined part that prompts the AI system into a specific persona, task and form for instance. And this is just a simple example. Unlike some other opinions, prompt engineering is not going away. It is the language in which we communicate with AI systems, and it will create results based on our input.  Understanding Prompting*, Prompt Engineering** and Pre-Prompts*** doesn’t require to learn a new language. It simply helps enormously to make it very clear what you want. And that is not only with Artificially Intelligent Beings but also with Naturally Intelligent Beings like you and me 🙂

Three areas of AI-aided Innovation

During an innovation process, Innovation Teams are working on thousands of different things. This makes innovation a bit difficult. It is very interesting to realize, we can group almost all those task into three innovation relevant action groups:
RESEARCH — VALIDATION — IDEATION

ChatGPT Prompting for Innovation Teams for RESEARCH

Before you start solving a problem and trying to develop a disruptive solution,  you will want to research the problem you are trying to solve and understand it from every aspect.
The old way of doing research is to go into a search engine, look for background information, scientific papers, research documents, user reviews, make some sentiment analysis in the social web, and so forth. Then after reading through all the documents you summarize key findings, prioritize certain findings, and craft your assessment. It takes at least several days if not weeks. In a typical innovation process we find at least 14 areas, where you need some research – not only in the beginning. All in all, it is approximately 2-3 months of uninterrupted work, to do all of your research with your team.
The new way of doing research is to ask questions to a Large Language Model like ChatGPT and get the answer immediately. You may want to consider some 30 minutes to finetune your question and still validate the outcome. But 14 research occurrences, times 1 hour is two days and not two months. You just collapsed your innovation research by 95% (2 days instead of 40 days). Even if you need to work on it for a few weeks to get savvy – the gain in efficiency is phenomenal. Also a great example for how AI is not replacing people – but hyper efficient people using AI will replace people who don’t.

To improve it further, you can optimize not only the research itself but the preparation from research reports. Tell the AI to serve the findings in nice tables, even do the priorities for you. It helps to create so called PRE-PROMPTS to save yourself time for the next research. The more time you invest in pre-prompts the better the outcome.
When you are looking for an innovation opportunity in your business which should be your first research you may use this pre-prompt:

THE COMPLETE PROMT:
Hey, act as my researcher for the automotive industry
After framing the research I will ask you a specific question.
Help me identify the biggest challenges electric vehicle manufacturer are facing.
Electric vehicle vendors seam to have trouble selling their cars.
A well known problem seams to be the still rather spotty charging station network.
My specific question is:

Now here comes your actual quetion which can obviously vary by a large degree:
“what are the unmet expectations from divers that are not met by sales, marketing and service teams of EV manufacturer?”

Now here you have a post-prompt to define the output format and maybe a follow on question:

Create a list of up to 25 of such expectations in the order of priority to fix.
Furthermore

“Create me a list of European electro vehicle companies with a specific challenges and what challenges they have.”

You see you can get as extensive and complex as you like.  Now, you may argue this is still rather manual and a maybe a bit painful to find all the right questions and variations, output forms and so forth. Correct, but as long as you do this manually for yourself, it is OK and a great learning experience.

In a system Like BlueCallom this is all done for you.  As we know the deeper meaning of each question and how we want to feed the KPI framework, we created all the Pro-Prompts and Post-Prompts for you. Meaning that your research time collapses even further down to minutes.

ChatGPT Prompting for Innovation Teams in IDEATION

Now – Ideation is a completely different beast. Ideation is your very personal creative process. But we have to say, after studying idea composition from the neuroscience angle, we can even drive am AI to be creative. Let’s say you want to invent a way to overcome the physical limits of todays geothermal energy generation. You find that you just can’t drill deeper than 12 KM. And you want to match the 500°C of a Nuclear Power Plant to drive a steam engine, and you need to get 25 km deep into the crust to reach that temperature (allmost anywhere on the planet) and you can’t pump water much higher than 100 meter – let alone 25 km. That is a serious problem because you are hitting the limit of physics. Get creative with the prompt too and frame it well. 🙂

THE COMPLETE PROMT:
Act as my universal expert and inventor in geothermal energy
Here is my challenge: I want to get to the average temperature of 500°C in the earths crust but drilling directly into the earth, the current limits are at 12 km depth.
I want to use the geothermal power and drive a steam engine in the gigawatt range.
I don’t care how the path to the 500°C is formed or found and how the steam reaches the turbine or the turbine gets to the steam.
Think of solutions that may not be drilling, may not be directly vertically, maybe longer than 25 km and so forth.
What solution can you develop?

The result is not yet optimal but it definitely stimulate the human brain beyond normal stimulation based on widening the conditions. This is by the way how we reached the conclusion to drill in a different angle and build the solution as it is shown as one of the Concept Innovations.

ChatGPT Prompting for Innovation Teams in VALIDATION

Yet another category of tasks or questions are about Opinions or Decisions during the innovation process and their respective validations. End here again, it is a good idea to start with prompt-framing. Let the AI understand your opinion or decision and find arguments for it or against it. Assuming you are in the early phase of your innovation and you want to define the audience that you like to involve in your innovation process so you get timely feedback and win some early supporter.

THE COMPLETE PROMT:
Act as my innovation advisor for our automotive business.
Here is the scenario we work in: We are developing a disruptive automobile that can change its shape and how it is utilized with a push of a button.
We need fans, influencers and highly committed car owners for testing validation and priming the market.
We thing about selecting our largest customers, like the biggest fleet managers, the biggest taxi organizations, and the biggest rental car businesses.
We expect the initial selection will also be the first buyer and get our disruptive product instantly into the market.
Here is the help I want from you: Can you confirm that this is the best possible selection? Do you see any risks in this selection? If this is not the best selection, what would you suggest? If yes, why do you agree with my selection? What other advice would you give me?

This prompt again the importance of the prompt-framing as you will get different responses with different pre prompts. If you would add even more details the result would differ again. You will see if you add your company size and the industry of your clients, the price range of your products, the better the answers are getting.

BlueCallom ChatGPT Integration

The ChatGPT integration in BlueCallom is taking it a few steps further. As we come partly from the neuroscience part of the world, we can shape prompt-framing in a way that idea process relevant considerations can work in the background. Also we are now working on the integration into the KPI-Framework is a way that the human rating of the answer can become a relevant factor of the success predictability.

With this integration, BlueCallom is a multi-dimensional AI application where we use six different AI systems for very different jobs such as
1) AI aided neuro ideation
2) innovation research
3) success prediction
4) content visualization
5) user guidance and administration
6) decision validation support (new)
With the fusion of neuroscience and AI, BlueCallom gained an unparalleled competitive advantage. Time to Innovation has basically collapsed from years to weeks. Innovation success predictability is reaching single digit percent accuracy, and innovation team process administration almost entirely vanished away through our AI-driven guidance system.

You will soon be able to experience the full integration with BlueCallom in the new BlueCallom Lite version.
You can get on the list now as we will roll it out sequentially based on demand and support resources.

[su_button url=”/bluecallom-lite-invite/” target=”blank” background=”#ffffff” color=”#007abb” size=”7″ center=”yes” radius=”5″ icon_color=”#f8f8f8″ text_shadow=”0px 0px 0px #000000″ class=”font-weight: 900″]Put me on the BlueCallom Lite list (it’s free)[/su_button]

 

References
* Prompting is the art of writing a request for an AI system. It maybe as simple as how “How much is an average salary for sales people in France.

* Prompt Engineering, is the ability to create detailed background information adding context relevant information.
Or it maybe highly specific like : “I want you to act as my  personnel consultant and help me establish terms for attracting exceptionally successful sales people. Help me establish attractive terms for exceptional sales people in the medical instrument business in Paris, France with 2-4 years of sales experience. Generate a list of things and a salary to offer. Additionally add a few tips that candidates may find very compelling to work for us.

*Prompt-Framing, is the technique to frame a prompt with the ideal amount of background data (not too much not to scarce), that can be provided as a pre-prompt. Followed by a rather simple prompt with the core of needs to be dealt with, and the AI can return an answer that is in context of the pre-prompt. The post-prompt may then provide further instructions in natural language, such as output format, follow on questions and more.

0 replies

Leave a Reply

Want to join the discussion?
Feel free to contribute!

Leave a Reply