Instructing Computers

Telling a Computer What to Do

• Natural Langugage
• Applications
• Programming

Natural Language

• Comfortable for humans
• Lacks specificity
• Lacks precision
• Difficult for computers to interpret perfectly

Ambiguous Language

The professor said on Monday he would give an exam.

Ambiguous Language

One morning, I shot an elephant in my pajamas. How he got in my pajamas I don’t know.

—Groucho Marx

Large Language Models

• Completions are inherently stochastic
• Results may not be exactly correct or what a user intends
• Need precise inputs to match precise requirements

Large Language Models

• Incredibly powerful tools
• Growing more powerful rapidly
• Still require expert knowledge for many use cases

Apps

• Another approach to instructing computers
• Work well for pre-defined tasks
• Must be created by someone

Programming

• Provides direct instruction to the computer
• Can be used to compute anything computable
• This can do nearly anything

Programming

• Not just for experts
• Can be a satisfying, creative experience
• Opens up new means for intellectual expression
• Is basically a superpower

Programming Challenges

• Learning a new language
• Understanding what computers can do efficiently
• Solving problems algorithmically

Problem Solving

• Required to program computers
• Analysis at appropriate level of abstraction

Computational Thinking

Formulating problems such that solutions can be represented as computational steps

Process

• Problem Formulation - Conceptualize a problem
• Solution Expression - Formulate unambiguous algorithm
• Execution - Computer runs instructions to produce result
• Evaluation - User explores result to confirm thinking

Algorithm

• Step-by-step process to solve a problem

Example Problems

Simple Computer

• Tape-based computer with programmable head
• The number under the head can be scanned
• One number can be stored in the head
• A counter in the head can be incremented

Instructions

• Move head left
• Move head right
• Stop if scanned is empty
• Stop if scanned matches stored
• Increment counter
• Increment counter if stored matches scanned
• Repeat

Count the number of items on the tape

counter should be equal to the count of the items on the tape when the machine stops

Counting

1. Stop if scanned is empty
2. Increment counter
3. Repeat

Search through list for an item matching stored

The machine head should stop with the head on the first match if there is a match or stop after the end of the list if there is no match.

Search

1. Stop if scanned matches stored
2. Stop if scanned is empty
3. Move head right
4. Repeat

Count the number of matching items in a list

counter should be equal to the count of matching items on the tape when the machine stops

Counting Matches

1. Increment counter if scanned matches stored
2. Move head right
3. Stop if scanned is empty
4. Repeat