Local AI firm makes waves globally

KARACHI: Not all programmers who create cutting-edge artificial intelligence (AI) products that are used widely around the world are from Silicon Valley.

Some of these nerds may be found in Gulistan-e-Jauhar, a neighbourhood that typifies middle-class Karachi. Even though they attended regional universities, they prevailed over Ivy League graduates on the AI hill.

Since its debut earlier this month, AIAV, a personalization engine developed by technology consulting company AlphaVenture, has been riding the AI wave.

The application gives ChatGPT, a generative large language model (LLM), a "context layer" to assist users in creating customised raps, bios, and dating profiles based on their social network persona.

Its inventors assert that it became "one of the fastest generative AI products" to reach more than 200,000 user signups from 194 countries in just three days, and the buzz on Twitter supports this claim.The amount of computational resources needed for generative AI based on LLMs, such as ChatGPT and Google NLP, is very large. A context engine created by AIAV extends their functionalities. It instantly recognises pertinent information, like one's Twitter feed, for example. The LLMs are then fed this data in order to operate more effectively.

Many people appear to believe ChatGPT is an authority figure. Saad Mughal, chief technology officer of AlphaVenture, claims that is untrue.

Your Twitter feed, for instance, can be flooded with political hot takes. However, requesting ChatGPT to create a brief essay on your political views can be fruitless. such's because ChatGPT doesn't have such granular access to open data.One requires a "context layer" — an AI engine with "thousands of lines" of code built by Mr Mughal and Hammad Khan, CEO of AlphaVenture — to obtain such kinds of AI-based insights.

You next enter your Twitter handle—or anyone else's, for that matter—on the "Twitter look-alike bot" and wait a little while for the AI engine to go through all of your tweets. Your timeline is filtered by its context layer to find the tweets that best represent you, and it then delivers the chosen information to ChatGPT along with targeted questions. The outcome appears shortly after in the form of a dating profile, personalised rap, and brief biography.

The AI engine is more than just a game. Given that the corporation is already looking for ways to monetize it, it promises its inventors a large sum of money.

According to user feedback, the rap feature—personalized compositions that are ready to be written and sung—is very well-liked.

It's extremely likely that people will collaborate with AI-based tools that create music and turn their rap into genuine songs. The concept of working with dating apps who might want to incorporate the AI-generated dating profile feature on their platforms is also viable.

The context engine is still prone to mistakes, it must be said. For instance, an AI-generated biography of the former finance minister Miftah Ismail claimed he had studied electrical engineering at California State University, even though he hadn't, and that he had a master's degree in the subject from Stanford University. 

According to Mr. Mughal, these are still early-stage issues that will be resolved as the tool develops. The level of the engine's accuracy will also be raised by employing more LLMs in addition to ChatGPT, he continued.