Hi guys! It's being a long time since my last post. But, recently a friend of mine, Thiago (follow and read his posts too ;D ), that recently began writing, brought me some inspiration to start writing again.
While I write this, I'm listening to: YearMix 2020 - Progressive House by Miss Monique on Youtube.
A bit of history
Since ChatGPT was introduced in 2022, there's nothing else that the people are talking about. I know that OpenAI brought a lot of technology that no one was expecting at that moment.
Maybe all of the 'big techs' like Google, Microsoft, X, and others were scared because they didn't have a ready product to compete. Even if they owned a lot of data to make one, or they owned enough technology, OpenAI came out ahead.
Let's talk about coding
Since Chat GPT was released, I began using it. Sometime later, when Github Copilot was released, I became one of the paying users. AI is part of my developer tools.
A couple of times ago, with a client of mine, I created an integration with the Open AI Api. They already owned a chat between employees and consumers, even a chat with a local bot that following some rules, has the capacity to talk with consumers.
So, I connected AI to this chat bot and thing works like a charm. With my capability to work with AI, I solved some problems and brought some reliability to the chat bot.
My first "all in AI job"
After all of that context... Some days ago, a company contacted me, they was needing a service to read some uploaded files related with the company, extract some information and finally convert into a content that people can easily understand.
AI INPUT
The first thing about AI, is that AI is not "All Powerful Omniscient Machine". You need to prepare the input data. It means that you have to comprehend the data provided from the company and the AI needed input data.
Another important aspect to worry about is the financial issues. You can't create an AI integration without worrying about how much money that integration will cost to the company. Remember that they will pay to each character (or for a certain amount of characters), then the input is very, very important to work in.
AI OUTPUT
The second thing is that, as the input, comprehend the AI output and the company expectations. We can't resolve every human problem with AI, then we need to comprehend what AI can solve and equate the expectation of the company. Then, expectations equalized, we can work on the AI output and create the final product.
FINAL PRODUCT
AI can solve a lot of different problems in our world. Therefore, the final product can be totally different for each AI integration. You have to take some tech decisions with the company: "It's really necessary send data to AI each time it was requested? Can we cache the answers to use later?". And things like that.
That's all, Devs
In conclusion, working with AI involves finding a careful balance between understanding its capabilities and aligning with client expectations.
You need to make the client informed about the data input needing, about the cost (speccially) and the possibilities of AI.
I hope I continue working with AI and writing about it to brought some persperctive to all of AI enthusiasts.
Top comments (2)
A very interesting story + tutorial to start working with AI.
Thanks bro!