AI and investments: Where is the capital influx directed?

Investments in AI: The future of investments in Artificial Intelligence is looking diverse with room for capital influx in AI infrastructure, research and development and foray into foreign languages, according to the CEO of AI notetaker Fireflies.ai. 

While 2023 was a year of experimentation for Large Language Models (LLM) and Generative AI, 2024 will be more about market consolidation, according to Krish Ramineni, founder and CEO of Fireflies.ai. 

“A lot of investments are going towards R&D, specifically around large language models (LLMs). We're doing more work on foreign languages and supporting foreign language support as well,” Ramineni said.

Fireflies.ai, which is based in California, has also been working towards automated technology like detection, and developing smarter AI with each interaction.

“From what we found from conversation with video conferencing players and through our own data, Fireflies is the most popular AI notetaker on all the major video conferencing platforms,” he said.

AI advancements are set to be on a path to be less of a software and more like teammates or colleagues, he said.

Investments Incline

Ramineni added that there is also a new trend of building AI companies with or without capital.

“It is hard to pick winners (in AI), since the market is still young. While Perplexity has investments from multiple players (including Amazon’s Jeff Bezos), Midjourney on the other hand is not raising capital and remaining bootstrapped,” Ramineni noted.

He also cited examples of infrastructural investments in Ola’s Krutim and Sarvam AI, which also has partnerships with Microsoft for voice-based AI applications. 

Ramineni highlighted how investors are grappling with unique value, as many applications are merely ‘GPT wrappers’ that provide a purpose-built user interface (UI) to interact with OpenAI's GPT models.

Venture capitalists investing into the advancing technology, however, need to realise not everything will help them gain returns, he said.

Other players in the AI meeting transcription space include Otter, and Google’s own workplace tool called Duet AI. Microsoft also enables transcription from pre-recorded audio.

Ethical Guardrails

But as AI goes for market consolidation, there is also a need for ethical guardrails around AI, he said, since it can be used for misinterpretation through fake news and videos, creating chaos.

It is easier to fake something as it becomes more believable, making it harder to distinguish between what is real versus what is artificially generated, he added.

For this, Fireflies makes sure anything that AI generates, says or answers can be traced back to the transcript to attribute it to a source, he said.

This measure is likely for AI to not insert interpretations on its own.

“And then other investments for us, it doesn't directly maybe relate to ethical AI, but more around corporate governance, security, and ensuring enterprises have full control over their data and making sure that data is not trained and that their information is not being put into (open source) models,” he added.

He also addressed the debate of self-governance versus government regulations in AI, saying that “self-governance will be faster to streamline, since the government can stall innovation.”

Recently, Adobe, Google, Meta, Microsoft, OpenAI, Tiktok and others joined hands to work on guardrails for AI during elections. Meta has also rolled out an India-specific Elections Operations Center for the responsible use of generative technologies.