1
00:00:02,480 --> 00:00:03,040
Hi guys.

2
00:00:03,040 --> 00:00:04,040
Welcome to this session.

3
00:00:04,040 --> 00:00:10,080
So in this session we'll talk about prompt engineering understanding prompt engineering in detail how

4
00:00:10,080 --> 00:00:12,800
this actually works on the ChatGPT tool.

5
00:00:13,160 --> 00:00:15,200
So what is prompt engineering.

6
00:00:15,200 --> 00:00:17,640
So let's read through this and understand it clearly.

7
00:00:17,880 --> 00:00:24,960
Prompt engineering is a process of designing and optimizing prompts used in natural language processing

8
00:00:25,280 --> 00:00:28,320
models, such as ChatGPT or virtual assistants.

9
00:00:28,760 --> 00:00:35,760
This involves crafting prompts that are clear, concise, and effective in eliciting the desired result.

10
00:00:36,040 --> 00:00:43,080
So, for example, prompt engineering is like making an effective fishing lure just as well designed

11
00:00:43,120 --> 00:00:49,560
lure is more likely to catch the fish, a well crafted prompt will also more likely to give us the desired

12
00:00:49,560 --> 00:00:50,240
results.

13
00:00:50,920 --> 00:00:56,080
There are three main principles of prompt engineering, which you keep in mind while you are working

14
00:00:56,080 --> 00:00:56,960
with this tool.

15
00:00:57,160 --> 00:00:59,160
The first can be being specific.

16
00:00:59,560 --> 00:01:05,600
The more criteria you give, the more focused the output will be.

17
00:01:05,600 --> 00:01:09,630
The more specific information we are going to provide to the ChatGPT tool.

18
00:01:09,870 --> 00:01:17,430
It is going to give us much more properly desired structured responses based on that work in steps.

19
00:01:17,430 --> 00:01:23,070
So we have to break down a task into smaller chunks of tasks which we give it.

20
00:01:23,110 --> 00:01:29,670
We can't go and ask ChatGPT to write a book for us, so we have to structure it down into small, small

21
00:01:29,670 --> 00:01:30,390
steps.

22
00:01:30,390 --> 00:01:36,070
So maybe discussing about the topic of the book, which what will the topic of the book?

23
00:01:36,350 --> 00:01:39,230
Then thinking about the table of content, what will be the topics?

24
00:01:39,230 --> 00:01:41,910
Chapter one, chapter two, what will be the chapters for it?

25
00:01:42,790 --> 00:01:45,590
And then working on each of the chapters one after the other.

26
00:01:46,710 --> 00:01:51,870
So working in steps really helps to get a much more better responses from the.

27
00:01:52,910 --> 00:01:56,470
The other thing which you can keep in mind is iterate and improve.

28
00:01:56,670 --> 00:02:02,510
So once you get a response from ChatGPT, we can rework on the on the inputs as well, which we are

29
00:02:02,510 --> 00:02:03,030
giving.

30
00:02:03,230 --> 00:02:07,350
Plus we can improve on the outputs which ChatGPT is providing us.

31
00:02:07,550 --> 00:02:09,670
So we can go ahead and modify that.

32
00:02:09,670 --> 00:02:11,350
We can ask in a different manner.

33
00:02:11,470 --> 00:02:17,220
We can bring out different versions of the output which we have got and ask again to re improvise on

34
00:02:17,260 --> 00:02:17,660
that.

35
00:02:17,660 --> 00:02:20,820
So all those things has to be a continuous process.

36
00:02:21,380 --> 00:02:26,900
So this is how your prompt engineering is going to evolve and improve over a period of time.

37
00:02:27,420 --> 00:02:30,220
Now what makes a good prompt.

38
00:02:31,380 --> 00:02:35,860
Great prompts all come down to the data the model was trained on.

39
00:02:35,860 --> 00:02:39,940
So ChatGPT data which is at the back end which they have been pulling up.

40
00:02:40,100 --> 00:02:43,260
It's all based on the data they have pulled up.

41
00:02:43,260 --> 00:02:47,100
And now based on that it's giving us the responses, its parameters.

42
00:02:47,140 --> 00:02:48,260
Good prompting.

43
00:02:48,580 --> 00:02:54,460
Since we can only control one of these, here's what the good prompting looks like.

44
00:02:55,340 --> 00:03:01,860
So the good prompting we have to keep in mind clear and concise language that is direct and unambiguous.

45
00:03:01,900 --> 00:03:07,580
Okay, whatever prompt you are giving to the tool has to be very clear and concise to the point.

46
00:03:07,900 --> 00:03:13,020
Vague prompts will produce vague responses.

47
00:03:13,300 --> 00:03:16,060
So we just need to go ahead and keep that in mind.

48
00:03:16,100 --> 00:03:21,780
The persona that you are assigned to ChatGPT, also known as who it will be acting as in the prompt.

49
00:03:22,140 --> 00:03:24,160
Okay, so there can be one aspect of it.

50
00:03:24,200 --> 00:03:27,400
We'll talk about it as well where you can act.

51
00:03:27,440 --> 00:03:33,680
You can ask ChatGPT to act in a certain manner, like a philosopher, maybe a doctor or an engineer.

52
00:03:33,960 --> 00:03:39,200
Okay, so in that manner, you can ask ChatGPT to act in a certain manner and give the responses.

53
00:03:39,680 --> 00:03:44,000
The other thing is the information and the examples that you provide, also known as your input.

54
00:03:44,040 --> 00:03:51,240
The more examples, specific information you're going to give in your input, the responses will be

55
00:03:51,520 --> 00:03:54,280
that very well high quality responses.

56
00:03:54,280 --> 00:04:00,320
You will get a specific task that you are requesting ChatGPT to complete, also known as the desired

57
00:04:00,360 --> 00:04:01,040
output.

58
00:04:01,400 --> 00:04:01,800
The.

59
00:04:02,240 --> 00:04:08,360
We have to make sure that the we have to ask for a specific task, so that only then we can expect to

60
00:04:08,360 --> 00:04:10,280
get a desired result out of it.

61
00:04:10,320 --> 00:04:11,880
Refinement as needed.

62
00:04:11,880 --> 00:04:18,120
Once you receive your first response, also known as reiteration, until receiving the desired output.

63
00:04:18,160 --> 00:04:23,360
So this is again refinement of the outputs which you are getting and again asking in a different manner

64
00:04:23,360 --> 00:04:25,640
to get better outcomes from ChatGPT.

65
00:04:26,520 --> 00:04:28,840
Now main prompting steps.

66
00:04:28,880 --> 00:04:33,910
What you can keep in mind is defining the problem or goal in a clear manner.

67
00:04:33,910 --> 00:04:34,270
Clear.

68
00:04:34,270 --> 00:04:40,270
Articulate what you want ChatGPT to help you with using relevant keywords and phrases in the prompt,

69
00:04:40,270 --> 00:04:47,150
you need to input the right most useful industry and topic related terms into the prompt to get the

70
00:04:47,150 --> 00:04:48,230
desired result.

71
00:04:48,510 --> 00:04:55,310
Write the prompt so crafting a concise prompt that clearly communicates the information and task that

72
00:04:55,310 --> 00:04:57,950
is required to be performed by the tool.

73
00:04:58,390 --> 00:05:06,310
Also, apart from this testing, your evaluation process iteration process has to be a part of it.

74
00:05:06,350 --> 00:05:08,190
Generate responses with ChatGPT.

75
00:05:08,230 --> 00:05:11,510
Once you get the responses, you evaluate the results.

76
00:05:11,510 --> 00:05:15,430
You go ahead and iterate on it and ask for a improved.

77
00:05:15,670 --> 00:05:20,150
You modify that and ask in a different manner to ChatGPT to get the desired responses.

78
00:05:20,470 --> 00:05:27,150
So this is what is going to be prompt engineering how you give your prompt to the tool, which is going

79
00:05:27,150 --> 00:05:30,030
to define the kind of responses you will get from it.

80
00:05:30,350 --> 00:05:31,350
I hope this makes sense.

81
00:05:31,350 --> 00:05:33,390
You understand prompt engineering now.

82
00:05:33,510 --> 00:05:37,990
Thank you so much guys for listening to this and I will see you in the next video.