Once men turned their thinking over to machines in the hope that this would set them free. But that only permitted other men with machines to enslave them. – Frank Herbert, Dune
ChatGPT v3 became publicly available November 30, 2021. More than 1 million people used it in the first week. You enter in a question or request, ChatGPT responds with beautifully written and substantial answers, and retains the context of your ‘chat’ so you can follow-up. ChatGPT creates text articles, can write poems and songs in a style you desire, and can produce working code in multiple computer languages.
I asked ChatGPT to describe itself to me. I entered “Please describe ChatGPT and its potential impact in 200 words or less.” It gave me this response in 17 seconds:
ChatGPT is a large language model developed by OpenAI. It is trained using a variant of the transformer architecture and is capable of generating human-like text on a wide range of topics. It is a powerful tool for natural language processing and can be used in a variety of applications, such as language translation, text summarization, and conversation generation.
One of the most significant potential impacts of ChatGPT is in the field of natural language understanding and generation. With its ability to generate human-like text, ChatGPT can be used to create more natural and engaging interactions between humans and machines. This could be used to improve customer service and support, and to create more realistic virtual assistants and chatbots. Additionally, ChatGPT could be used to improve the efficiency and effectiveness of natural language search and information retrieval systems.
Another potential impact of ChatGPT is in the field of content creation. With its ability to generate text on a wide range of topics, ChatGPT can be used to generate news articles, stories, and other written content. This could be used to save time and resources for content creators, such as journalists and writers.
Overall, ChatGPT is a powerful tool that has the potential to revolutionize the way we interact with machines, as well as how we create and consume written content.
I asked ChatGPT to write a haiku about encouragement:
Encouragement bloom,
Like spring flowers after snow,
Hope in every word.
I asked ChatGPT to give me 5 blog post ideas for advanced strategic decision-making, then asked for an outline of #5:
- “5 Common Mistakes in Strategic Decision-Making and How to Avoid Them”
- “The Role of Data Analysis in Strategic Decision-Making”
- “Using Scenario Planning to Anticipate Future Challenges in Strategic Decision-Making”
- “The Importance of Encouraging Creativity and Innovation in Strategic Decision-Making”
- “Strategic Decision-Making in the Digital Age: How Technology is Changing the Game”
Please give me an outline for #5
I. Introduction
- Briefly introduce the topic of strategic decision-making in the digital age
- Explain how technology is affecting the way decisions are made in organizations
II. The impact of big data on strategic decision-making
- Discuss the importance of data collection and analysis in decision-making
- Describe the challenges of managing big data and how they affect decision-making
III. The use of artificial intelligence in strategic decision-making
- Explain how AI can be used to improve decision-making processes
- Discuss the challenges of implementing AI in decision-making and how to overcome them
IV. The role of digital platforms in strategic decision-making
- Discuss the impact of digital platforms on the way decisions are made
- Explain how organizations can use digital platforms to gain a competitive advantage in decision-making
V. The future of strategic decision-making
- Describe the emerging technologies that will shape the future of decision-making
- Discuss how organizations can prepare for the future of strategic decision-making
VI. Conclusion
- Summarize the main points discussed in the article
- Offer recommendations for organizations looking to improve their strategic decision-making in the digital age.
All this in seconds, in English language queries, using a web browser with an internet connection.
I could go on to share how ChatGPT responded to these queries:
- Write a 10-minute homily on the Beatitudes
- Summarize the key events in the first year of the French Revolution
- Write a poem about snow fences in the style of Mary Oliver
- Suggest a meal plan for one week for a low carb but not keto diet, but without avocadoes
- What do you recommend for a tourist see in Rome that hardly anyone visits?
- Generate an Excel formula to strip out all characters before the @ symbol in column C, then merge with the text in columns D and F, putting the answer in a new column.
Here was the answer to the last request – note that it explains the formula and recommends how to adjust it!
=CONCATENATE(RIGHT(C1,LEN(C1)-FIND(“@”,C1)),D1,F1)
This formula uses the CONCATENATE function to join together the text from three different cells: the part of the text in column C after the “@” symbol, the text in column D, and the text in column F. It uses the RIGHT function to get the right side of the text in column C after the @ symbol, which is determined by the subtraction of the number of characters before the “@” symbol using the LEN function and the position of the “@” symbol using the FIND function. The result will be displayed in a new column. Please note that you’ll need to adjust the cell references (C1, D1 and F1) to match your specific sheet.
ChatGPT, along with related AI tools for image and video generation, are a step change improvement over previous interactive capability. This is a game-changer level capability comparable to the first electronic spreadsheet, internet search engines, and online GPS-driven maps with real-time driving instructions.
It’s astounding, frankly.
The creators at OpenAI are using user feedback to make it better. ChatGPT v4 coming in 2023 will be good enough to pass a bar exam and certain medical certifications. This kind of tool will accelerate our need to re-engineer education approaches. Not all but much of the hype is warranted.
A digest of my thoughts and observations:
- ChatGPT is a tool, which can be used for better and worse. We shape our tools and our tools shape us. Humans don’t uninvent tools. This is a new kind of tool in our world and we’ll need to learn to use it well.
- This is a category of digital capabilities, not just one tool. This kind of AI tool will wind up in many other tools. It’s already being embedded in apps for writing, image and video composition.
- It’s crucial that we understand the limitations of these tools. ChatGPT responses are perfectly written, but sometimes wrong. (It reminds me of some people who are perpetually 100% confident even if they’re wrong.) Authority + Assertiveness + Inconsistent Accuracy is a dangerous combination.
- This capability is sufficiently advanced that it will accelerate our conversation about how to educate children and youth, what constitutes original work, how we manage information, and many work roles (not plumbers, but certainly coders and copywriters).
- This level of AI, as it evolves into new applications, could become a trillion dollar economic value and disrupt multiple business models, including ad-supported search engines.
- ChatGPT can’t think for you, but it can support better thinking. It can’t make a decision for you, but it could support better decision-making. ChatGPT takes nothing away from your responsibilities, but could improve your ability to fulfill your potential. This AI capability will continue to improve and will pressure us to focus on generating better questions and using our imaginations.
- It’s a mistake to humanize this tool concept. ChatGPT is not conscious. Nor will it somehow mysteriously develop a consciousness. It reflects and relies upon its designers and the data sets used to train it. God/The Muse/The Voice does not talk with algorithms and robots.
ChatGPT is at the ‘inflated expectations’ stage of the Gartner hype curve, but I suspect not yet at the peak:
Neil Postman’s questions for a new technology (from his lecture On Culture’s Surrender to Technology) are useful here:
1. What is the problem to which technology claims to be a solution?
2. Whose problem is it?
3. What new problems will be created because of solving an old one?
4. Which people and institutions will be most harmed?
5. What changes in language are being promoted?
6. What shifts in economic and political power are likely to result?
7. What alternative media might be made from a technology?
I’ll leave these as thought problems for you, and share more thoughts and ideas below:
ChatGPT Limitations
An acquaintance, a former special forces operator, tells me that he was trained to excel at finding vulnerabilities, exploiting them, and blowing stuff up. In that spirit, what are the limitations of this awesome AI technology?
- It requires massive compute power, electricity, connectivity, and digital devices.
- It can’t do anything in the physical realm.
- It can write a beautiful homily, but it can’t “rejoice with those who rejoice and mourn with those who mourn.” It has no capacity to hear God and respond. It has no capacity to tailor a sermon or guidance to what one congregation needs to hear on this particular Sunday.
- It can craft a legal brief, modeled after precedent, but it has no ability to make a judgment.
- It’s only as good as the data sets used to train it. It has no ability to discern what is truthful and beautiful within that data set.
- It doesn’t cite its sources for what it produces. (This gives it a veneer of originality, but remember, it’s fundamentally a clever mimic based on pre-existing information available to it.)
Business Perspective
OpenAI was started with seed money from Elon Musk, Microsoft, and others. ChatGPT runs on Microsoft Azure cloud computing, not Amazon or Google’s cloud platforms. Microsoft has offered to buy 49% of OpenAI, which would give OpenAI a market valuation of $29B. Tools like ChatGPT integrated into Bing search and Office could reposition Microsoft relative to Google and Apple. (I worked closely with Microsoft for several years earlier in my career – they’re savvy, think long-term, and have massive cashflows.)
ChatGPT is currently (largely) free to use in research mode, but it’s not free to run. It requires massive compute power to run large language models and train them. ChatGPT already has an API, so it’s straightforward to connect the capability to other software. Presumably licensing the capability to others (“Just add our AI to your system for $X per month for up to 20,000 queries”) is how OpenAI will generate revenue.
The business model is crucial for AI success. Most people don’t understand that Alexa and Siri are money losing operations. Amazon lost about $10B in 2022 with Alexa devices and running the back end software. ChatGPT-level sophistication is impressive, but organizations can’t lose money forever.
We know that government agencies interface with technology companies. I expect ongoing controversies about how much governments (foreign and domestic) influence the businesses operating this technology.
Open questions to be explored
Amidst all the hype, there are uncomfortable questions which don’t have answers today. Examples:
Who owns the copyrights to what ChatGPT produces? If I ask ChatGPT to create a collection of haiku poems, can I publish them as “my” work? Will OpenAI get licensing revenue if a musician uses ChatGPT to create lyrics to what becomes a hit song? [Same questions for AI tools that create image and video.] If something profitable emerges because ChatGPT created a derivative from copyrighted work, should the original author receive compensation?
What governance models should apply to ChatGPT? Is OpenAI liable for text it generates which someone finds offensive? (Side note: Microsoft had to shut down an earlier and impressive chatbot called Tay because it essentially became a Nazi afficionado.) This is a powerful tool, and will certainly be useful to propogandists and marketers alike. There should be conversations, too, about the training data sets – what content is selected, and how are they weighted?
How does AI fit into quality education? This AI capability will impact our education models. How can we preserve the value of the difficulty of writing when there’s a cheap tool that could easily get you a B- on your next essay assignment? The process of writing and editing is a mind-sharpening effort, with long-term payoffs. And how will we evaluate teachers and professors in the academic world if it’s easy to craft decent text articles that sound authoritative even if they’re unoriginal?
To what extent will people choose to interact with an AI like ChatGPT rather than interacting with other people? What would be the secondary consequences of that choice? We know this about ourselves: We desperately long for communication and social interaction, and need it, but also fear and dread it because it’s difficult and fraught with challenges. We’re less than one lifetime into a digital communication age and already struggling with some of the downsides. This level of AI toolset introduces new complexities.
What other questions come to your mind?
Near-term Technology Trends
ChatGPT-type of AI tools are already being integrated into software for writing (Lex), research, and managing personal information collections (e.g., Notion). That trend will accelerate now. Search is next – we’ll likely see a combination of a list of links to content and an AI-generated commentary.
Expect that biometrics will be much more important for information security. A simple password will not be sufficient to protect your information. The typical Captcha step will easily be overrun by AI.
Plagiarism-checkers will need a new evolution.
Before long we’re going to have to testify that our work is original and not created by an AI. That’s going to push enhanced trust-checking software, probably more biometrics.
Using ChatGPT Wisely as a Tool
New tools give us options. I suggest that AI tools like ChatGPT put a premium on good questions, imagination, and discernment. ChatGPT will be a boon for generating ideas and starting points for further editing. One can get a pre-first draft easily, and use that as inspiration and mechanical help.
Treat the output like you would a wet bar of soap in the shower – don’t grip too tightly or you lose it. Do not put all your trust in an AI tool. Keep your ‘Captain Skeptical’ hat close by, and check what it generates carefully. This is a powerful research tool to explore topics and ideas. If you work with code, it could reduce carpal tunnel problems.
We must be ever vigilant and discerning as these tools evolve and improve. The responses are slick and impressive, even if they’re wrong. There will be times when you should sniff out garbage amidst a lovely meal presentation. The way to become discerning about lies and manipulation is to focus your attention on what is true and beautiful – so that the lies and manipulation stand out as wrong, even if you’re not sure why or how. Years ago, banks trained employees handling money to identify counterfeit bills by only letting them handle genuine currency bills for weeks. After that, they could rapidly pull out bills that didn’t feel right somehow and set them aside. Nearly all of these were indeed counterfeits. A wise person saturates himself with the truth.
Be mindful of ChatGPT’s limitations (see list above) and don’t over-humanize these tools. Remind yourself that AI is all A and no I. Honor the value of the tools without diminishing the value of flesh-and-blood people who are both physical and spiritual in ways no AI can ever attain.
I can’t easily predict what will happen in education and fields of work which are heavily text oriented. I can’t easily predict what new legal and regulatory changes are coming. I can say that we’re going to need wise and patient people in this effort. This technology is advancing far faster than our education, social, and legal systems can evolve. We should not trust these serious decisions with infantilized unserious ‘adults.’
As one of my mentors advises, use tools without being a victim or addicted to them, and sleep with one eye open.
