Amr Awadallah on AI propaganda
From Amr Awadallah, CEO of Vectara and co-founder of Cloudera, discusses the future of AI search · · PeopleReign
“I'm actually very worried about that problem I don't have an answer for what it's go the proper way for handling would be one idea I have which can be true but might not it requires research to prove out that it can work out is we are going to need the antivirus of facts like we have viruses for software viruses for software happens all the time even though it's illegal by the way if you create a virus for software and the federal government catches you you are going to prison but we still have viruses happening left and right because we live in a global world and we live in a world where not everybody is a nice person but the solution was we're going to have regulations that prevent these things that's step number one which is great but two we're going to have technology that catches these things right so I think we need a the technology that is an anti-propaganda technology where if I'm hearing a message is talking to me right now trying to convince me with something a number will show up in the upper right corner these statements coming out of his mouth right now with a 70% probability they're being said to manipulate you they're not being said to truly educate you right can we have something like that we I think we need that like the future will need that will require that so maybe that's a startup idea for somebody to go work go work on.”
On , Amr Awadallah, Cofounder at Cloudera, spoke about AI propaganda during Amr Awadallah, CEO of Vectara and co-founder of Cloudera, discusses the future of AI search on PeopleReign.
About Amr Awadallah
Amr Awadallah, co-founder of Cloudera and CEO of Vectara, has been active in discussions about enterprise AI adoption, focusing on the challenges of accuracy, security, and explainability. He has stated that an estimated 95% of enterprise AI projects fail, attributing this to a lack of clear strategy and the difficulty of moving from prototypes to production. Awadallah has argued that retrieval-augmented generation (RAG) is not dead, contrary to some claims, and that even with larger context windows, RAG remains essential for grounding AI responses in factual data. He has emphasized that hallucinations are an inherent problem in large language models, citing Vectara's hallucination leaderboard which shows even the best models hallucinate around 1% of the time, and has advocated for a "guardian agent" system to monitor outputs. Awadallah has also commented on industry dynamics, describing the release of DeepSeek-R1 as a "Sputnik moment" for the U.S. that disproves the assumption of a two-year lead over China. He has expressed concern about the competitive pressure this places on companies like OpenAI and Nvidia. On the topic of open source, he has noted that while it helps with developer awareness, it is difficult to monetize at scale, drawing on his experience at Cloudera. Awadallah has advised aspiring entrepreneurs to focus on solving a real problem rather than being driven by technology or the desire to get rich, and has stressed that passion for the work is a key factor in long-term success.