# Trajectory's Ronak Malde says AI should learn with humans in the loop

Markdown copy of https://realalpha.app/news/trajectorys-ronak-malde-says-ai-should-learn-with-humans-in-the-loop/ on Real Alpha, the best minutes from long interviews with the people building AI. Cite and link the page, not this copy.

[News](https://realalpha.app/news/)· Training Data

By [Rushabh Sheth](https://realalpha.app/news/authors/rushabh-sheth/) Nov. 5, 2026, 9:05 a.m. ET

Ronak Malde, co-founder of Trajectory, said AI that keeps learning should stay tied to people. He said so on the podcast Training Data, in an episode posted Thursday.

## Why you'd care

This matters because AI is moving into more of the tools people use at work and at home. How those systems learn decides who steers them. Malde thinks people should stay in charge of that.

Malde contrasted two research paths. One is AI helping to build better AI, which researchers call recursive self-improvement. He described it as putting a model in a black box and making it the best AI researcher itself. He called that "a scary world" because it raises the question of whether the AI will keep human interests in mind.

He prefers continual learning, where an AI system learns from its experience working with people. He said the answer is to build it so "it keeps humans in the loop." He said he doesn't want agents, meaning AI that keeps working on a task by itself, going rogue.

## In the news

The topic is in the news. [Tekedia](https://www.tekedia.com/anthropic-warns-it-may-be-approaching-recursive-ai-self-improvement-in-revised-sec-filing/) reported on Nov. 4 that Anthropic revised its SEC filing. In it, Anthropic warned that it may be approaching a stage where AI systems speed up the making of more capable models.

Malde runs a company built around continual learning, so he has a stake in that approach. He didn't say which path will win.

Watch the 4 minutes where he explains the difference. [20:11](https://realalpha.app/v/32vxW6GMeLY/#t=1211)

The moment

[Bringing Continual Learning to Every Company: Trajectory’s Ronak Malde](https://realalpha.app/s/continual-learning-keeps-humans-in-the-loop/#t=1445), Training Data, Nov. 5, 2026, from 20:11 to 24:19

The story

[Anthropic Warns It May Be Approaching Recursive AI Self-Improvement in Revised SEC Filing](https://www.tekedia.com/anthropic-warns-it-may-be-approaching-recursive-ai-self-improvement-in-revised-sec-filing/), Tekedia, by Samuel Nwite, Nov. 4, 2026

Quote freely with a link ([how to cite us](https://realalpha.app/press/)). Something wrong? Write to [hello@realalpha.app](mailto:hello@realalpha.app) and we'll correct it ([how we pick and check items](https://realalpha.app/news/about/)).

[Ronak Malde Co-founder, Trajectory](https://realalpha.app/builders/ronak-malde/)

From Bringing Continual Learning to Every Company: Trajectory’s Ronak Malde, on Training Data

## More news

[All news](https://realalpha.app/news/)

1. **[Jassy says AWS's 2027 is largely sold out despite $220 billion capex](https://realalpha.app/news/jassy-aws-2027-largely-sold-out-220-billion-capex/)**
   a16z Podcast · Nov. 5
2. **[fal's Gorkem Yurtseven says image models beat a price war](https://realalpha.app/news/fals-gorkem-yurtseven-says-image-models-beat-a-price-war/)**
   Sourcery · Nov. 5
3. **[Alexander Wissner-Gross says AI can now edit other AI directly](https://realalpha.app/news/alexander-wissner-gross-says-ai-can-now-edit-other-ai-directly/)**
   Moonshots · Nov. 4