AI Engineering

[AshishKumar1]

AI has become an essential part of human life nowadays. From gathering details about any topic to its implementation. Almost many things are revolving around AI. Whether we need to know anything, write a code logic or learn some new method or knowing about how to implement it, everything can be done by AI.

Artificial Intelligence became the burning topic around 2022 when Chat GPT developed by open AI gained its popularity among people. Artificial Intelligence is basically development of computer system which can perform some of the task that require human intelligence like recognizing speech, answering to our question etc.

Advancement of AI to GEN AI made it a hot topic of discussion among people. There are other AI bot as well apart from Chat GPT which is BARD developed by google and BING which was developed by Microsoft. Open AI brought 2 version of Chat GPT one is GPT 3.5 which is free for everyone and the other is GPT Plus also known as GPT 4 which is a subscription-based service provided by open AI. In Chat GPT the GPT stands for Generative pretrained transformer. Chat GPT is based on Large Language Model (LLM). LLM is basically a machine learning models which can perform natural language processing tasks as Machine learning models that are build and optimized to understand and process the text which can be understood by humans and then models can perform tasks with that text. LLM are basically trained to calculate probability which can suggest words based on prior words. So basically, the whole idea behind the LLM is to get some input and then to produce some output so as to complete the sequence of words. The sequence can be of anything like sequence of number or any other sequence.

Some of the difference between GPT 3.5 and GPT 4 are.

• GPT 3.5 Produces result quickly as compared to GPT 4
• GPT 3.5 has less reasoning capability as compared to GPT 4 and this also results in a reason why GPT 4 is bit slower than GPT 3.5
• GPT 4 has stronger conciseness as Compared to GPT 3.5 and here conciseness and we simply summarize as efficiency.

Some of the limitation of Chat GPT are.

• It has limited training data as it only has data limit until September 2021 only. Unless we have access to Chat GPT plus or GPT 4.
• Chat GPT result works on pattern not logic like it predicts works and it assign words on basis of probability.

Now we can use these AI tools efficiently by using prompt engineering technique like

• We can setup custom Instruction for Chat GPT where we can provide it certain information of ours so that it considers those information while generating those results.

• We should keep our message concise and should include specific points to generate more accurate prompt. We can do this by following Role + context + Goal technique here Role basically means like we should specify our role while writing prompt like if we are developer then we should say like I am developer and by the context it means that we should be specific and concise to the main context and Goal means like we should be specific to our goal like what we are expecting rather than deviating from it.

We can get the result of our question by providing the template as well like in the format which we want it to generate results.
Chat GPT can generate Image by using DALL-E and, we can upload an Image and can ask it to summarize the content of the Image. We can use the feature of code interpreter so as we get more specific solution to our query related to codes. One of the feature of code interpreter is that it has the OCR which is Optical character Recognition. We should follow the iterative process until the program is done like suppose we developed a program and then we executes it and we encounter that certain features has been missed so in that case we can ask it again to fix those issues or error. Following this process until the goal is achieved is termed as following iterative process until the goal is achieved.

We can use Chat GPT for Iterative development by adding more and more features by splitting our requests across multiple prompts, Chat GPT can also be used for debugging by reporting error and bug and then we can ask the possible solution to fix it, Chat GPT can be used for refactoring like we can use Chat GPT to refactor our code to get improvements ideas.

So these are some features which we can use to leverage AI in our day to day activity and to boost our efficiency. Hope this helps.