They had a Copilot in the movie Airplane.

Well, the naughty autopilot’s name was Otto, only revealed in the trailing credits. The copilot was supposedly Roger Murdoch, but really Kareem Abdul-Jabbar. For what it is worth, I’m pretty sure I have internalized all the jokes in “Airplane!” (note the exclamation mark).

Roger Roger

They had a Copilot in the movie Airplane.

Well, the naughty autopilot’s name was Otto, only revealed in the trailing credits. The copilot was supposedly Roger Murdoch, but really Kareem Abdul-Jabbar. For what it is worth, I’m pretty sure I have internalized all the jokes in “Airplane!” (note the exclamation mark).

Roger Roger

and Roger the cabin boy?

I've had some numerical classes generated by AI with bad numerics. I recently asked a question for research and it just gave me some code which I then decided to use but the AI had used the Big Ball of Mud anti-pattern so I manually separated out 3 or 4 classes. I've since wondered whether I could have got Copilot, which I've never used, to do the refactoring for me and, if it had generated the code, whether the code would have been just the one class.

Ugh 

Well, the naughty autopilot’s name was Otto, only revealed in the trailing credits. The copilot was supposedly Roger Murdoch, but really Kareem Abdul-Jabbar. For what it is worth, I’m pretty sure I have internalized all the jokes in “Airplane!” (note the exclamation mark).

Roger Roger

and Roger the cabin boy?

No, you should definitely watch the movie!

I got bored half-way through, unfortunately.

What's all about, Alfie?

I got bored half-way through, unfortunately.

That’s deeply disturbing

I got bored half-way through, unfortunately.

That’s deeply disturbing

More bad news, 

Fountainhead
Hitchhiker's 

also bored me to hell.

In fairness, I avoided Evelyn Waugh. He was a silly man.

Never seen Fountainhead, and it was never a good idea to try to explicitly visualize the fantasies of Douglas Adams, since it was bound to lose out to the images we’d already generated in our heads when reading the books.

I remember watching THG as a torture. DAs writing is all about literary tricks: irony, absurd, slightly philosophical satire, linguistic sense and logic used only to break them... The writer/narrator is a character in the book too. It just won’t work in a movie, which is about picture and acting, and the story must be relatively simple.

Could be interesting. I reckon Microsoft could come up wit AI-enabled design patterns or something? I've seen it a bt with C#.

Prompt engineering probably needs Ontology

1. the branch of metaphysics dealing with the nature of being.
2. a set of concepts and categories in a subject area or domain that shows their properties and the relations between them.

But MIA when computers got faster and ANNs were rediscovered.

Just 25 years ago intelligent (mobile) agents and KQML were all the rage. They fell out of fashion.
And probably a bridge too far for non-academic applications.

I'm fairly sure that a chunk of requirements analysis could be "AI-automated". And not necessarily just code.

I would say many business opportunities.

Could be interesting. I reckon Microsoft could come up wit AI-enabled design patterns or something? I've seen it a bt with C#.

Just brainstorming here:

1. Take all internal chats, documents, emails and meeting transcripts loosely connected with a topic.
2. Feed it into an LLM with a prompt asking it to spot budding product ideas and output a list with descriptions and links to sources.
3. Triage the list in a smaller group.

Yes, it's not rocket science, it's all been done before, albeit with human stakeholders (see Sommerville and Sawyer (1997)). It's called Requiremenents Engineering but you need a special breed of s/w folk with stamina for it! (*)
And Ontology!

(*) the s/w industry has been stuck in syntax wars for the last 30 years.

Question , has ChatGPT long-term memory? 
If you can do it manuaily, then "AI" as well?? Analogous reasoning and all that.

A domain architecture (Duffy, 2004) is a reference model for a range of applications that share similar structure, functionality and behaviour. A domain architecture is a kind of meta model that can be seen as a template for more specific systems.
We discuss five basic forms and one ‘composite’ form. Thus, in total we have six domain categories:
MIS (Management Information Systems): Produce high-level and consolidated decision-support data based on transaction data from various sources.
PCS (Process Control Systems): Monitor and control values of variables that must satisfy certain constraints. We are primarily interested in exceptional events  that must be handled in the software system.
RAT (Resource Allocation and Tracking) systems: Monitor a request or some other entity in a system. The request is registered, resources are assigned to it, and its status in time and space is tracked. This is probably the most common model that we encounter in applications and it could be the starting point for many medium-sized applications.
MAN (Manufacturing) systems: Create finished products and services from raw materials and data.
ACS (Access Control Systems): Allow access to passive objects from active subjects. These are similar to security systems and to the Reference Model in large computer systems.
LCM (Lifecycle Model): A composite model that describes the full lifecycle of an entity; it is a composition of MAN, RAT and MIS models.
 

It does have memory: https://openai.com/index/memory-and-new ... r-chatgpt/

Sounds like an add-on/fix?

Today's AI is not what it used to be, for example

Winston, P. H. (1980) Learning and reasoning by analogy. Communications of the ACM 
23(12) 689–703.

example
Is a MIS system for hours planning the same A-Z as a system for system resources in DEC VAX/VMS computer environments? Large environments
Answer is yes, grosso modo (real-life projects). In  a week will have a prototype up and running.