Download as pptx, pdf, or txt
Download as pptx, pdf, or txt
You are on page 1of 15

THE AI OPPORTUNITY

AND
PROMPT ENGINEERING
THE AI
Hardware Services Software
3,000

OPPORTUNITY
• Automated systems 2,000

• Data-Driven insights

USD Billion
• Customer service enhancement
• Personalized experiences 1,000
• Healthcare advancements
• Creative applications

0
The forecasted market size
$2745B

21

22

23

24

25

26

27

28

29

30

31

32
for AI in 2032 (in USD):

20

20

20

20

20

20

20

20
20

20

20

20
THE NEED FOR ENGINEERING PROMPTS

GARBAGE GARBAGE
IN Large Language Model OUT
PROMPT ENGINEERING PRINCIPLES

Write clear and specific Give the model time to


instructions "think"

Use delimiters Specify steps

Ask for structured output Instruct to work out its own


solution before rushing to a
“Few-shot” prompting conclusion
PROMPTING PRINCIPLE - 1

Write clear and specific instructions

Reduced chance of irrelevant or incorrect answers.

A clear prompt is not the same as short one.


WRITE CLEAR AND SPECIFIC INSTRUCTIONS

Use delimiters to indicate distinct


parts of the input, such as:
• triple quotes (”””)
• triple backticks (```)
WRITE CLEAR AND SPECIFIC INSTRUCTIONS

Ask for structured output such as:


• in form of steps
• as an email/message
WRITE CLEAR AND SPECIFIC INSTRUCTIONS

"Few-shot" prompting:
Give successful examples of
completing task, then ask
model to perform the task.
PROMPTING PRINCIPLE - 2

Give the model time to think

Logical steps to get the answer.

The model might guess.

More model thinking means increased computational


effort on the task.
WRITE CLEAR AND SPECIFIC INSTRUCTIONS

Specify the steps required to


complete a task
WRITE CLEAR AND SPECIFIC INSTRUCTIONS

Instruct the model to work out its


own solution before rushing to a
conclusion
MODEL LIMITATION

Hallucination

Model makes up false information or facts

It could make statements that sound plausible


but are not true
ITERATIVE PROMPT DEVELOPMENT

ERROR
ANALYS
IS

RESPONS
EFROM IDE
THE A
MODEL

IMPLEMENTATI
ON
(PROPMTING)
USE CASES

Summarize text Transforming data

Draw inferences Expanding data

You might also like