Is Krutrim Based on GPT-3?

Shortly after the launch of Krutrim’s beta chatbot, it has found itself mired in controversy.  A particular Reddit user’s post has cast a long shadow over the authenticity of Krutrim’s AI technology. This inquiry has led to a broader conversation about innovation in the Indian technology landscape.

The Reddit Revelation

It started with a Reddit user with the handle ‘V1p3rspeaks’ probing Krutrim’s chatbot, seeking to uncover the technology underpinning its AI. The response—or lack thereof—ignited suspicion and eventually led to the revelation that the chatbot might be operating on a technology not of the company’s own making. This discovery has since raised the fundamental question: Is Krutrim’s AI simply a rebranded version of OpenAI‘s GPT-3?

The Reddit thread paints a grim picture of disillusionment. Users have been unsparing in their critique, likening Krutrim to past ventures by Aggarwal that drew heavily from international models.

It’s crucial to note that when others attempted to probe Krutrim AI with the same line of questioning about its foundational technology later, the results were inconsistent. While Reddit user V1p3rspeaks uncovered a direct admission of the chatbot’s reliance on OpenAI’s GPT-3 model, subsequent inquiries by different users did not always yield the same transparent response. This inconsistency has added another layer of complexity to the debate. It suggests that either Krutrim AI’s responses are being fine-tuned to avoid such disclosures, or there is a variance in how the chatbot processes and responds to similar queries. This discrepancy is particularly noteworthy and has further fueled speculations surrounding the true extent of originality in Krutrim AI’s technology.

Krutrim AI’s Meteoric Rise to Unicorn Status

Krutrim, the brainchild of Bhavish Aggarwal, became India’s fastest startup to achieve unicorn status, claiming a valuation of $1 billion. With a $50 million investment led by Matrix Partners India (also an investor in Ola), Krutrim’s ambitious goal is to build a large language model that leverages local Indian languages alongside English. The vision is grand—a voice-enabled conversational AI assistant fluent in multiple Indian dialects.

Aggarwal’s venture is not shy about its objectives. It aims to bring a beta version of its chatbot to consumers, followed by APIs for developers and enterprises. There is also talk of developing in-house capabilities to manufacture AI-optimized chips, positioning Krutrim as a potential leader in AI computing stacks.

Investor confidence in Krutrim is evident from the hefty valuation, driven by the global race to back AI innovation. Investors are betting big on AI’s transformative potential across industries, with Krutrim positioned as a hopeful frontrunner out of India.

Despite India’s robust startup environment, the nation has yet to significantly impact the global AI race. With tech giants like OpenAI and Google leading the charge, Indian ventures like Krutrim, Reliance’s yet-to-launch AI, and other startups like Sarvam are under immense pressure to deliver something comparable, if not superior. QX Lab AI’s recently launched Ask QX also supports 12 Indian languages and is vying for a share of the same pie.

The financial stakes in developing a large language model are astronomical, with companies like OpenAI and Anthropic raising billions to fund their endeavors. This context makes Krutrim’s funding achievements noteworthy but will only intensify the scrutiny that it will have to contend with. Bhavish Aggarwal’s popularity and the success of Ola will only add to the scrutiny.

A Crossroad for Krutrim

As Krutrim navigates these turbulent waters, it stands at a critical juncture.

The AI community and potential consumers await a clear response from the company. Can Krutrim substantiate its claims of innovation and independence from existing models like GPT-3? The answer to this question will not only define Krutrim’s future but also set a precedent for what innovation means in the Indian startup landscape.

The company’s assertions of creating a “complete AI computing stack” are impressive. Aggarwal’s vision for Krutrim, to spearhead India’s venture into AI, reflects a bold and ambitious roadmap. The startup’s plans to develop voice-enabled conversational AI and in-house AI-optimized chips showcase a commitment to creating a comprehensive AI infrastructure.

However, the path to such innovation is fraught with challenges, not least of which is the looming shadow of already-established AI behemoths. With behemoth investments in the sector, the pressure is on new entrants like Krutrim to not just emulate but transcend the capabilities of these existing models.

Weighing the Scale of Innovation

The disparity in innovation between developing a new model from the ground up and adapting an existing one is significant. The Reddit user’s findings have brought this issue to the forefront, emphasizing the need for startups to pursue genuine innovation rather than repurposing existing technologies. As the conversation unfolds, the Indian tech community is becoming increasingly vocal about demanding not just localized solutions but also technological breakthroughs that can compete on a global stage.

The Road Ahead for Indian AI

Krutrim’s journey exemplifies the broader context of India’s ambitions in the AI sector. With a burgeoning startup ecosystem, there is immense potential for India to make its mark. However, this will require a shift from following global trends to setting them, from playing catch-up to leading the pack.

Investors have shown they are willing to pour funds into promising AI ventures. Still, for these investments to bear fruit, startups like Krutrim must navigate the complex terrain of innovation with transparency and originality.

The Verdict Still Out

With the beta launch of its chatbot, Krutrim is under the microscope. The allegations stemming from the Reddit user’s findings have not been conclusively proven, but they have undoubtedly tainted the waters of Krutrim’s debut. For Krutrim to sustain its unicorn status and justify its valuation, it must unequivocally demonstrate that it is not just riding on the coattails of GPT-3, but is indeed a beacon of Indian ingenuity.

As it stands, the Indian tech community and global observers are left with a pressing question: Is Krutrim a genuine innovator, or is it leaning on the laurels of established AI giants like GPT-3? The answer will have far-reaching implications for Krutrim’s credibility, the perception of Indian startups, and the future of AI innovation in India. Only time will reveal whether Krutrim can dispel these doubts and emerge as a true pioneer in generative AI.