How I Query an Ai Engine, and How I Should be Queried by Ai?

It was during my online class at IAA (Indian Author Academy) that I came to know of ChatGPT ai. As part of the curriculum, we were asked to use this highly advanced tool of technology. It was still under development and learning through the prompts that users input.

We were almost spoon-fed initially with some warning about not to give out sensitive information and that the engine is capable enough to answer queries but at the same time could be inaccurate too! This part of it I did not get it.

My experience with Ai Querying

But then, while experiment with festival greetings image, I understood it. When I queried for image of Diwali greeting, certain logical function is slightly off-beat. Once the update happens, or/and large-scale change is done to be deemed version 5 of the engine. Then most of these issues will be addressed and redressed.

For instance, such as, say putting image of people in sinkable area almost walking on water. As per current reality, humans cannot walk on water unless they are highly evolved saints or yogis. That differential thinking is still lack in the ai.

Text-Based Queries Workable

The output when provided for a text-based query does not include the source from where these inputs are taken. Citation of the actual system of research text that was referred to in producing the output, is lacking in ai. When in the absence of source code or archives approached and distilled for information is not clear. If that be the case, then the accuracy of the output is in doubt.

But then, I use a free version so maybe in paid version, they might let us know from where the engine has been taking information. My assumption is that the information provided as from all those free to access blogsite or channels like Gutenberg and other outlet. If that be the case, then citing them would only add accuracy to the output text.

An Example: In Daily Publishing Industry

In the daily newspaper if the reporter were to use ai to search and collate material for article then if the distilled summary that is provided by the ai would be helpful. But is it true enough to print? That’s the burning question of the moment.

Citing sources and origin of the research made by ai, like from where the information is taken. This would make the reporter of the said scenario to make an educated and studied decision on the structure of his write-up.

Citing Methods Should be Interactive

The citing of written material should base a trigger and response condition using authentication key code which is passed between the source computer and Ai engine. Now this allow the output to be rechecked. When the ai is used in the intranet networking system or the internet networking system, the verification of the information collected should be done using different methods.

Verification symbols with Green-based colour coded images. — **Image:1** Verification Symbols in Green colour gradient

Popularity Cannot Be Termed as Authentication

If the site has more traffic or comments these cannot be used an indicator of authentic. The madness of research might become the reality for ai and so cause the breakout of hallucination of the engine.

Citing for the Intranet and the Internet

Intranet and Internet with iCloud connected universal view. — Intranet and Internet a pictorial view

When dealing with citing in the Intranet would be make a simple query and response sequence. Then, verification indicator based on unique code that is generated over the sequence of facts verifying checklist can be created and each machine should be uniquely recognised to allow the search and find of the erroneous areas and isolate to allow authenticity a space in the iCloud.

Verified Document after all the intranet portal's approval — **Image 2:** Verified Intranet Document

Verification Elements for the Intranet

When creating the Intranet each machine is by default uniquely identified as the Network Administration or Engineer would setup these machines to have access to the main server. The main server maybe a collection of linked up standalones forming the intranet within a single organisation maybe distributed across continents.

The act of colour coding the accuracy of the information or data can become a way to set specific colour for closely accurate to far from accurate colouring. This then will be checking method of the ai engine to find out if this is an authentic space where I am picking up information. So, the green and red spectrum of colours could be used for encoding the input access for the output of ai.

Verification Elements for the Internet

Verified Document in the Internet iCloud this has verification symbols from multiple platform — **Image 3:** Verification for Document that serve Authentication of the Verity

Internet is more of a global phenomenon and citing for across continents should carry the first occurrence of the research text with maybe the countries flag. Each country is responsible for the data from their servers of information dissemination.

The flag plus the few other colour codes for accuracy and agency from where this information can be cross-referenced for accuracy are some of the options that could be worked around would be a proper portals handshake and handing over verity of the data. All these are done in the behind the scenes by the computer. A final global verification for accuracy of data can be provided by an indicator or stamp that is visible to the user.

But For Now!

But for now, I am happy with the little surprises that ChatGPT pushes forward to me in my circle of interest that is English Language based Literature.