Krutrim to launch AI chatbot app soon, says Bhavish Aggarwal; what to expect

Krutrim is gearing up to introduce its inaugural product – an innovative generative artificial intelligence (AI) chatbot application. The brainchild of startup founder Bhavish Aggarwal, the venture aims to revolutionise digital interactions with its cutting-edge technology. Aggarwal recently took to social media platform X (formerly Twitter) to unveil the imminent launch timeline of the maiden AI offering, offering glimpses of its capabilities through shared screenshots.

According to statements from the startup, all AI products set for release will possess the remarkable ability to comprehend India's diverse linguistic landscape, encompassing a staggering 22 scheduled languages. Additionally, these pioneering products will boast content-generation capabilities across eight languages.

Aggarwal's announcement on February 4 hinted at the final stages of testing for the @Krutrim app, scheduled for release in the upcoming week. Despite expressing satisfaction with the app's current state, Aggarwal pledged continuous enhancement of the AI models even post-launch.

In a series of posts, Aggarwal showcased the chatbot's versatile capabilities, illustrating its proficiency in handling basic inquiries and providing insightful suggestions. However, one particular exchange drew significant scrutiny from netizens.

Upon querying the Krutrim AI chatbot about India's status as a country prior to independence, the app's response triggered a wave of controversy. Asserting that India existed as a sovereign entity before the British Raj, the response sparked debate among users. Aggarwal juxtaposed this response with that of OpenAI's ChatGPT, which offered a differing perspective on India's historical context.

Critics accused the Krutrim AI chatbot of exhibiting signs of "AI hallucination," a phenomenon wherein large language models generate erroneous or fictitious information. Many users argued in favour of ChatGPT's response, citing its alignment with historical accuracy. Additionally, concerns were raised regarding a user interface flaw, notably the misalignment of the 'Send' icon with the text box.

Krutrim has disclosed details regarding its family of Language Model Models (LLMs), comprising the Krutrim base and Krutrim Pro variants, the latter touted as a multimodal foundational model. These models have undergone extensive training on a corpus exceeding two trillion Indic language tokens. While the startup has hinted at voice-enabled functionality for its generative AI apps, the inclusion of this feature at launch remains uncertain.

The company has also provided insights into its AI model's performance benchmarks, presenting a comparative analysis against Meta Llama 2 7B, a coding assistant model trained on over 500 billion tokens. Despite surpassing Meta Llama 2 7B across various benchmarks such as Arc, Hellaswag, and Copa, Krutrim's performance lagged on benchmarks like MMLU, Bigbench, and Lambada_open.

Also Read Elon Musk denies funding talks for his AI firm xAI amid speculation