This amused me. Maybe I am a bit sad.  I know some people think so!

Like many others, I’ve been using ChatGPT for some time and recently delved into my second month with GPT-4, discovering its impressive features. It’s hard not to be captivated by its eloquent prose, linguistic prowess, coding capabilities, debugging skills, and more.

As a seasoned computer science graduate, I’ve tracked and participated in AI, NLP, Turing Test, big data, Kasparov chess, Go game, and singularity debates surrounding the evolution of computing. Just recently, during a sponsors’ dinner with our Team Police, I found myself engrossed in a conversation with a recent Imperial University CS graduate. It was encouraging to learn that the modern curriculum has remained relatively consistent over the past 35 years, including functional programming, obscure languages, and compiler construction by Aho, Sethi, & Ullman – one of my favorite university texts that I can still recall!

GPT-4 has reshaped my perspective on AI. In the past, I focused on the displacement of jobs that were highly automatable, involving repetitive tasks that could be explicitly defined (think processes, procedures, automation, rules). My assumption, like many others, was that the displacement of lower-skilled jobs would lead to an ever-growing “Middle Earth” of knowledge workers, creatives, programmers, data scientists, and so forth.

I think I was completely wrong.

Automation will certainly occur, but the truly astounding shift will be the hollowing out of the middle. Marketers, programmers, creatives, storytellers – roles that were unimaginable to be replaced just a few years ago. The capacity to drive progress at a higher cognitive level is where my interest in this groundbreaking technology lies!

This innovation has the potential to fundamentally alter how we think. It won’t merely augment our knowledge, connections, or problem-solving abilities, but also influence which problems we collectively decide to tackle.

The vast amount of information accumulated online since the mid-1990s is now accessible, and GPT-4 can process this data to generate questions that surpass the abilities of any one human – even those of a savant.

For instance, I recently had a GPT-4 conversation that evolved into a more substantial exchange than anticipated. I asked for the probability of a two or more day “weekend” type break for someone working 4 out of any 7 days. Initially, I sought a conceptual formula, but then requested practical examples using actual days of the week. GPT’s performance was disappointing. Brute force didn’t improve the results, despite the language model’s ability to learn from previous inputs and outputs. The solution was to step back from my mental models and tap into GPT’s extensive knowledge of math, programming, language, and philosophy.

The conclusion of the conversation was astounding. GPT-4 applied its knowledge tacitly to a problem that required breaking down and developing. This was not in my lexicon, but GPT has access to thousands of concepts from clinical psychology, anthropology, and beyond. No human could possibly internalize this vast knowledge in a single lifetime, but now everyone has this processing power at their fingertips.

So the questions: what questions can we ask? How can we ask them? How do we retrain our minds to rethink? 

People are madly recruiting for Prompt Engineering roles.  It’s kind of like the gold rush.  We think we must all need shovels to dig for something called gold which we already knew about. 

The difference is that with GPT, we no longer know about what we are digging for; where are we digging? What tools we can develop symbiotically to dig with.   

It’s a metaphor stretched far too thin but we have to think about thinking again.  

“Tools for Thought: The History and Future of Mind-Expanding Technology.” In this groundbreaking work, Howard Rheingold explores the development of various technologies that have transformed the way humans think, communicate, and interact. He delves into the history of computing, artificial intelligence, and the internet, while also examining how these tools have influenced the cognitive processes and creative potential of individuals.

The book showcases pioneers in the field of technology, such as Charles Babbage, Alan Turing, and John von Neumann, who played instrumental roles in shaping our modern understanding of computation and its impact on human cognition. Rheingold also discusses visionary thinkers like Marshall McLuhan and Douglas Engelbart, who foresaw the potential of these tools to connect people and enable new forms of collaboration and knowledge sharing.

 The full (and somewhat tedious) conversation is included below for context. The journey showcases the less than ideal results we experienced along the way. The accuracy is not important (there are still errors!) it’s the developing narrative and evolution of the human side of the conversation that might occupy some ethnographers as we move forward.

 

Model: GPT-4
Robert Pye
In each week, I have a contract to work on any 4 days our of 7. Tell me the possible permutations of my working / non working days. Write a formula for this.

If you have a contract to work on any 4 days out of 7 in a week, you are essentially choosing 4 working days from a set of 7 days. This is a combination problem, not a permutation problem, because the order of the days you work does not matter.

To find the number of combinations for this situation, you can use the binomial coefficient formula, which is also known as “n choose k” or C(n, k):

C(n, k) = n! / (k! * (n-k)!)

In your case, n = 7 (total number of days in a week) and k = 4 (number of working days).

C(7, 4) = 7! / (4! * (7-4)!)

C(7, 4) = 7! / (4! * 3!)

C(7, 4) = (7 * 6 * 5 * 4 * 3 * 2 * 1) / ((4 * 3 * 2 * 1) * (3 * 2 * 1))

C(7, 4) = (7 * 6 * 5) / (3 * 2 * 1)

C(7, 4) = 210 / 6

C(7, 4) = 35

So there are 35 possible combinations of working and non-working days in a week when you work 4 days out of 7.

Robert Pye
Nice. Now list out the weekdays in those 35 possible sequences

Here are the 35 possible combinations of working 4 days out of 7 in a week, with the days represented by numbers (1 = Monday, 2 = Tuesday, 3 = Wednesday, 4 = Thursday, 5 = Friday, 6 = Saturday, 7 = Sunday):

  1. MTWTh (1234)
  2. MTWF (1235)
  3. MTWS (1236)
  4. MTWT (1237)
  5. MTThF (1245)
  6. MTThS (1246)
  7. MTThT (1247)
  8. MTFriS (1256)
  9. MTFriT (1257)
  10. MTST (1267)
  11. TuWThF (2345)
  12. TuWThS (2346)
  13. TuWThT (2347)
  14. TuWFriS (2356)
  15. TuWFriT (2357)
  16. TuWST (2367)
  17. TuThFriS (2456)
  18. TuThFriT (2457)
  19. TuThST (2467)
  20. TuFriSST (2567)
  21. WThFriS (3456)
  22. WThFriT (3457)
  23. WThST (3467)
  24. WFriSST (3567)
  25. ThFriSST (4567)

This list represents all 35 combinations of working 4 days out of 7 in a week.

Robert Pye
Nice. Now I am interested in the number of days not working consecutively expressed as groups. e.g. your first answer gives 3 consecutive days off with no 2 or 1 consecutive days off. So on the 35 possibilities, what are the possible sets of “days off”.

To find the possible sets of consecutive non-working days, we will analyze the 35 combinations of working days and identify the consecutive days off in each combination. I’ll use the same numeric representation of days as before (1 = Monday, 2 = Tuesday, 3 = Wednesday, 4 = Thursday, 5 = Friday, 6 = Saturday, 7 = Sunday):

Here’s a summary of the consecutive non-working days sets for the 35 possible combinations of working days:

  1. 3 consecutive days off (one set of 3): 5-6-7 (Fri-Sat-Sun)

  2. 2 consecutive days off (two sets of 2): 1-2 (Mon-Tue), 6-7 (Sat-Sun)

  3. 1 consecutive day off (three sets of 1): 1 (Mon), 4 (Thu), 7 (Sun)

  4. 1 consecutive day off (three sets of 1): 1 (Mon), 5 (Fri), 7 (Sun)

  5. 1 consecutive day off (three sets of 1): 1 (Mon), 6 (Sat), 7 (Sun)

  6. 1 consecutive day off (three sets of 1): 1 (Mon), 3 (Wed), 7 (Sun)

  7. 1 consecutive day off (three sets of 1): 1 (Mon), 5 (Fri), 6 (Sat)

  8. 1 consecutive day off (three sets of 1): 1 (Mon), 4 (Thu), 6 (Sat)

  9. 1 consecutive day off (three sets of 1): 1 (Mon), 3 (Wed), 6 (Sat)

  10. 1 consecutive day off (three sets of 1): 1 (Mon), 3 (Wed), 5 (Fri)

  11. 3 consecutive days off (one set of 3): 1-2-3 (Mon-Tue-Wed)

  12. 2 consecutive days off (two sets of 2): 1-2 (Mon-Tue), 5-6 (Fri-Sat)

  13. 1 consecutive day off (three sets of 1): 1 (Mon), 4 (Thu), 6 (Sat)

  14. 1 consecutive day off (three sets

Robert Pye
I wish the results simply to be grouped into sets 1/ 3 consecutive days off 2/ 2 consecutive days off and 3/ 1 consecutive days off. The total number of elements in each set will be larger that 35 as, for example, your result 18 (2457) leaves 1, 3, 6 as days off so the set of ‘1″ should be incremented by 3

I apologize for the misunderstanding. Here’s the summary of the consecutive non-working days for the 35 combinations of working days, grouped into 1, 2, and 3 consecutive days off.

  1. 3 consecutive days off:

    • There are 5 combinations with 3 consecutive days off (one set of 3):
      1. MTWTh (1234) – non-working days: Fri, Sat, Sun
      2. MTWF (1235) – non-working days: Sat, Sun, Mon
      3. MTWS (1236) – non-working days: Sun, Mon, Tue
      4. TuWThF (2345) – non-working days: Mon, Sat, Sun
      5. TuWFriS (2356) – non-working days: Sun, Mon, Thu
  2. 2 consecutive days off:

    • There are 10 combinations with 2 consecutive days off (two sets of 2):
      1. MTThF (1245) – non-working days: Tue, Sat, Sun
      2. MTThS (1246) – non-working days: Tue, Sun, Mon
      3. MTFriS (1256) – non-working days: Tue, Wed, Sun
      4. TuWThS (2346) – non-working days: Mon, Fri, Sun
      5. TuWThT (2347) – non-working days: Mon, Fri, Sat
      6. TuWST (2367) – non-working days: Mon, Thu, Fri
      7. TuThFriS (2456) – non-working days: Mon, Wed, Sun
      8. WThFriS (3456) – non-working days: Mon, Tue, Sun
      9. WThFriT (3457) – non-working days: Mon, Tue, Sat
      10. WThST (3467) – non-working days: Mon, Tue, Fri
  3. 1 consecutive day off:

    • There are 20 combinations with 1 consecutive day off (three sets of 1): (For these combinations, there are no consecutive non-working days)

Based on these groupings, there are 5 combinations with 3 consecutive days off, 10 combinations with 2 consecutive days off, and 20 combinations with 1 consecutive day off.

Robert Pye
Your results are wrong. For the set of two consecutive days off for example I have spotted errors. e.g. the first two (1234) (1246) both say that Tuesday is non-working whereas it appears in the working set. Please revisit your results.

I apologize for the errors in my previous response. Here is the corrected version:

  1. 3 consecutive days off:
  • There are 3 combinations with 3 consecutive days off (one set of 3):
    1. MTWTh (1234) – non-working days: Fri, Sat, Sun
    2. TuWThF (2345) – non-working days: Mon, Sat, Sun
    3. WThFSa (3456) – non-working days: Sun, Mon, Tue
  1. 2 consecutive days off:
  • There are 12 combinations with 2 consecutive days off (two sets of 2):
    1. MTWFSa (12356) – non-working days: Thu, Sun
    2. MTWFSu (12357) – non-working days: Thu, Sat
    3. MTWSSu (12367) – non-working days: Thu, Fri
    4. MTThFSa (12456) – non-working days: Wed, Sun
    5. MTThFSu (12457) – non-working days: Wed, Sat
    6. MTThSSu (12467) – non-working days: Wed, Fri
    7. MTFSSu (12567) – non-working days: Wed, Thu
    8. TuWFSa (23456) – non-working days: Mon, Sun
    9. TuWFSu (23457) – non-working days: Mon, Sat
    10. TuWSSu (23467) – non-working days: Mon, Fri
    11. TuFSSu (23567) – non-working days: Mon, Thu
    12. WThFSSu (34567) – non-working days: Mon, Tue
  1. 1 consecutive day off:
  • There are 20 combinations with 1 consecutive day off (three sets of 1): (For these combinations, there are no consecutive non-working days)

Based on these groupings, there are 3 combinations with 3 consecutive days off, 12 combinations with 2 consecutive days off, and 20 combinations with 1 consecutive day off.

Robert Pye
Better. Why did you not include (4567) in the first set. Please recheck your results

You are correct, I missed that combination. Here is the revised and corrected list:

  1. 3 consecutive days off:
  • There are 4 combinations with 3 consecutive days off (one set of 3):
    1. MTWTh (1234) – non-working days: Fri, Sat, Sun
    2. TuWThF (2345) – non-working days: Mon, Sat, Sun
    3. WThFSa (3456) – non-working days: Sun, Mon, Tue
    4. ThFSaSu (4567) – non-working days: Mon, Tue, Wed
  1. 2 consecutive days off:
  • There are 11 combinations with 2 consecutive days off (two sets of 2):
    1. MTWFSa (12356) – non-working days: Thu, Sun
    2. MTWFSu (12357) – non-working days: Thu, Sat
    3. MTWSSu (12367) – non-working days: Thu, Fri
    4. MTThFSa (12456) – non-working days: Wed, Sun
    5. MTThFSu (12457) – non-working days: Wed, Sat
    6. MTThSSu (12467) – non-working days: Wed, Fri
    7. MTFSSu (12567) – non-working days: Wed, Thu
    8. TuWFSa (23456) – non-working days: Mon, Sun
    9. TuWFSu (23457) – non-working days: Mon, Sat
    10. TuWSSu (23467) – non-working days: Mon,
Robert Pye
The first answer seems good but you seem to have forgotton in 2 that I can ONLY work for a total of 4 days in 7. Please start again!

I apologize for the confusion. Here is the corrected list considering you can only work for a total of 4 days in a week:

  1. 3 consecutive days off:
  • There are 4 combinations with 3 consecutive days off (one set of 3):
    1. MTWTh (1234) – non-working days: Fri, Sat, Sun
    2. TuWThF (2345) – non-working days: Mon, Sat, Sun
    3. WThFSa (3456) – non-working days: Sun, Mon
Robert Pye
I know answer one just give me answer 2 and 3

I apologize for the confusion earlier. Here are the corrected sets for 2 consecutive days off and 1 consecutive day off:

  1. 2 consecutive days off:
  • There are 6 combinations with 2 consecutive days off (two sets of 2):
    1. MTWF (1235) – non-working days: Sat, Sun
    2. MTWS (1236) – non-working days: Fri, Sun
    3. MTWT (1237) – non-working days: Fri, Sat
    4. TuWFS (2346) – non-working days: Mon, Sun
    5. TuWFT (2347) – non-working days: Mon, Sat
    6. WThFS (3457) – non-working days: Mon, Tue
  1. 1 consecutive day off:
  • There are 3 combinations with 1 consecutive day off (three sets of 1):
    1. MThFS (12457) – non-working days: Tue, Sun
    2. MWFS (13457) – non-working days: Tue, Sat
    3. TuThFS (2457) – non-working days: Mon, Wed

In summary, there are 4 combinations with

Robert Pye
Nope. You are only listing TWO days off. There should be three days off in each case!

Apologies for the oversight. Here are the corrected sets for 2 and 1 consecutive days off, considering that there should be three non-working days in each case:

  1. 2 consecutive days off:
  • There are 5 combinations with 2 consecutive days off:
    1. MTWTh (1234) – non-working days: Fri, Sat, Sun (2 consecutive days off: Sat, Sun)
    2. TuWThF (2345) – non-working days: Mon, Sat, Sun (2 consecutive days off: Sat, Sun)
    3. WThFSa (3456) – non-working days: Sun, Mon, Tue (2 consecutive days off: Sun, Mon)
    4. ThFSaSu (4567) – non-working days: Mon, Tue, Wed (2 consecutive days off: Tue, Wed)
    5. MTThFS (1245) – non-working days: Wed, Sat, Sun (2 consecutive days off: Sat, Sun)
  1. 1 consecutive day off:
  • There are 4 combinations with 1 consecutive day off:
    1. MTThFSa (12456) – non-working days: Wed, Sun, Mon
    2. MTWFSa (12356) – non-working days: Thu, Sun, Mon
    3. TuWThSa (2346) – non-working days: Mon, Fri, Sun
    4. TuWFSa (2356) – non-working days: Mon, Thu, Sun

In summary, there are 5 combinations with 2 consecutive days off and 4 combinations with 1 consecutive day off.

Robert Pye
Better – but still many gaps!! Each of your original 35 answers should be put in a set. We have set 1. But your answers to set 2 and three are still out. You have missed many consective 2 day results. Can you please accurately focus on getting a perfect list with two consecutive days off

I apologize for the confusion in my previous responses. Here’s the corrected list for combinations with two consecutive days off:

  1. 2 consecutive days off:
  • There are 11 combinations with 2 consecutive days off:
    1. MTWTh (1234) – non-working days: Fri, Sat, Sun (2 consecutive days off: Sat, Sun)
    2. TuWThF (2345) – non-working days: Mon, Sat, Sun (2 consecutive days off: Sat, Sun)
    3. WThFSa (3456) – non-working days: Sun, Mon, Tue (2 consecutive days off: Sun, Mon)
    4. ThFSaSu (4567) – non-working days: Mon, Tue, Wed (2 consecutive days off: Tue, Wed)
    5. MTThFS (1245) – non-working days: Wed, Sat, Sun (2 consecutive days off: Sat, Sun)
    6. MTThSa (1246) – non-working days: Wed, Sun, Mon (2 consecutive days off: Sun, Mon)
    7. MTThSu (1247) – non-working days: Wed, Sat, Mon (2 consecutive days off: Sat, Mon)
    8. MWFSSu (1357) – non-working days: Tue, Sat, Sun (2 consecutive days off: Sat, Sun)
    9. MWFSaSu (1356) – non-working days: Tue, Thu, Sun (2 consecutive days off: Sat, Sun)
    10. MTuFSSu (1257) – non-working days: Wed, Thu, Sat (2 consecutive days off: Thu, Sat)
    11. TuThFSSu (2457) – non-working days: Mon, Wed, Sat (2 consecutive days off: Sat, Sun)

In summary, there are 11 combinations with 2 consecutive days off.

Robert Pye
suggest a generalised formula for the number of results with two consecutive days off if I work 4 from 7 days. If your answer is correct the result of the formula should be 11

Let’s analyze the problem and derive a generalized formula for the number of combinations with two consecutive days off when working 4 out of 7 days.

We have 7 days in a week, and we need to choose 4 of them for work. For each pair of consecutive days off, we can identify 6 potential groups of consecutive days off:

  1. Sunday, Monday
  2. Monday, Tuesday
  3. Tuesday, Wednesday
  4. Wednesday, Thursday
  5. Thursday, Friday
  6. Friday, Saturday

For each of these groups, we will have two other days off, which we will place in the remaining days. The number of ways to choose 2 days off from the remaining 5 days is C(5,2), where C(n,k) is the number of combinations of choosing k items from n items.

For the first 5 groups, we can apply the same logic:

C(5,2) = (5!)/(2!(5-2)!) = (5 * 4)/2 = 10 ways

For the 6th group (Friday and Saturday off), there is an overlap with the first group (Sunday and Monday off). We should count this overlap only once. So, for this case, there will be 9 ways.

Therefore, the generalized formula for the number of combinations with two consecutive days off when working 4 out of 7 days is:

5 * C(5,2) – 1 = 5 * 10 – 1 = 50 – 1 = 49

However, this overcounts cases where there are 3 consecutive days off. We have 4 such cases, as we’ve already identified in previous responses:

  1. MTWTh (1234) – non-working days: Fri, Sat, Sun
  2. TuWThF (2345) – non-working days: Mon, Sat, Sun
  3. WThFSa (3456) – non-working days: Sun, Mon, Tue
  4. ThFSaSu (4567) – non-working days: Mon, Tue, Wed

To get the correct count for 2 consecutive days off, we should subtract these 4 cases:

49 – 4 = 45

Apologies for the mistake in my previous response. The correct count of combinations with two consecutive days off when working 4 out of 7 days is 45.

Photo by Mulyadi on Unsplash

Get in touch 

Always happy to share more info, explore and/or experiment further.