frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Open in hackernews

Tell HN: Perplexity is defaulting to religious sources for secular queries

1•fallinditch•1h ago
I recently ran into a bizarre retrieval pattern while on a Perplexity chat session that suggests their RAG might be heavily skewed or actively gamed by niche SEO.

I was researching a topic completely unrelated to religion (technical/cultural analysis) when the model provided a series of confident but obviously incorrect answers. When I audited the citations, 8 out of 10 sources were religious publications - think biblehub, think churchofjesuschrist, and various denominational blogs - none of which were relevant to the objective facts I was seeking.

It feels like the "Domain Authority" weights in their index are being exploited. I guess it's a niche for a few agencies now explicitly offering "Christian Perplexity SEO" [1] to help faith-based organizations "measure trustworthiness" in AI scanners.

Between that and the general push for "diversity of perspective" in LLM tuning, it seems secular, objective queries are being caught in the crossfire.

Has anyone else noticed a sudden influx of faith-based or think-tank sources in their AI Search results lately?

[1] https://himfirstmedia.com/perplexity-seo/

Comments

fallinditch•1h ago
--looking back at the chat session the religious sources appeared when I challenged Perplexity to ask it why it mislead me.

It then lied again when it denied that it had included religious sources.

A typical link from the sources is to Luke 21:8

"He replied: “Watch out that you are not deceived. For many will come in my name, claiming, ‘I am he,’ and, ‘The time is near.’ Do not follow them."

The War on Drugs Is Why Your Bus Never Showed Up

https://maxmautner.com/2026/01/17/bus-marijuana.html
1•mslate•3m ago•0 comments

The Case for a Product Manager's Personal Agent

https://substack.com/inbox/post/185033315
1•carni234•4m ago•0 comments

How to write a good spec for AI agents

https://addyosmani.com/blog/good-spec/
1•saikatsg•4m ago•0 comments

Reliability Engineering 0→1

https://medium.com/@sandhyavinjam/reliability-engineering-0-1-how-to-scale-from-startup-chaos-to-...
1•sandhyavinjam•5m ago•0 comments

The AI Engineer Roadmap

https://www.aihero.dev/ai-engineer-roadmap
1•saikatsg•12m ago•0 comments

Show HN: Automated Google Slide Reports for multiple entities

https://github.com/rvignesh89/gslides_automator
1•rvignesh•18m ago•0 comments

Show HN: Yashiki – A tiling window manager for macOS in Rust, inspired by River

https://github.com/typester/yashiki
2•typester•19m ago•1 comments

Show HN: Omelo- AI pet health companion, Health timelines and Daily care

1•animalcatcher•20m ago•0 comments

Show HN: Early web-inspired writing platform

https://writing.ink/
1•eamonobr•22m ago•0 comments

Farath's Bi-Weekly Code Security Brief – Jan 1-19 Vulns SAST/SCA/DAST Playbooks

https://farathappsec.substack.com/p/faraths-biweekly-code-security-brief
2•farathshba•25m ago•1 comments

Wdlite

https://wdlite.toolforge.org/?item=Q686797
2•altilunium•26m ago•0 comments

Corrupt pardons corrode trust: Coffeezilla – Voidzilla [video]

https://www.youtube.com/watch?v=0ydymkrshVc
1•lofihiphophop•29m ago•0 comments

A crash-resistant serial terminal for messy high-speed data

https://github.com/C3POAdmin/8N1Term
2•C3POAdmin•30m ago•0 comments

All is not well between Meta CEO Mark Zuckerberg and Meta's AI

https://timesofindia.indiatimes.com/technology/tech-news/all-is-not-well-between-meta-ceo-mark-zu...
1•nsoonhui•30m ago•0 comments

Hiring at India's Big Four outsourcers stalls, as AI seemingly makes an impact

https://www.theregister.com/2026/01/19/hcl_infosys_tcs_wipro_results/
2•Brajeshwar•31m ago•0 comments

Show HN: Jello, popular games with your own images and sounds

https://www.jello.app
2•postatic•31m ago•0 comments

Show HN: Online List Maker – simple, syncing lists built on Durable Objects

https://onlinelistmaker.com/
2•pw•36m ago•0 comments

Show HN: CodeAnswr – Stack Overflow alternative with E2E encryption

https://codeanswr.com
3•mobinpo•36m ago•1 comments

We used to celebrate science and innovation (2025)

https://www.freethink.com/the-material-world/techno-humanist-manifesto-chapter-9-section-1
1•Brajeshwar•38m ago•0 comments

Show HN: 100% Agentic AI Comedy Podcast

2•helsinki•47m ago•0 comments

MTV Rewind

https://wantmymtv.xyz/player.html
1•boilerupnc•49m ago•0 comments

I built a contract-aware invoice analysis tool

https://concilely.com
2•natedev•55m ago•1 comments

Is Bilt 2.0 worth it?

https://alexchao.substack.com/p/is-bilt-20-worth-it
1•alexchaomander•57m ago•1 comments

Bypassing Gemma and Qwen safety with raw strings

https://teendifferent.substack.com/p/apply_chat_template-is-the-safety
2•teendifferent•1h ago•1 comments

Thicc

https://github.com/elleryfamilia/thicc
1•handfuloflight•1h ago•0 comments

Show HN: Intent Layer: A context engineering skill for AI agents

https://www.railly.dev/blog/intent-layer/
2•Hunter17•1h ago•0 comments

Show HN: Pdfwithlove – PDF tools that run 100% locally (no uploads, no back end)

https://pdfwithlove.netlify.app
3•pratik227•1h ago•0 comments

Nano Banana Prompt Library – AI Image Prompts

https://nano-banana.app/nano-banana-prompts
3•leegrayson2023•1h ago•0 comments

If You Think This Instrument Is Hard to Play, Try Building One

https://www.nytimes.com/2026/01/14/arts/music/oboe-laubin-jim-phelan.html
2•tintinnabula•1h ago•1 comments

Engaging healthily with chess: an Acceptance and Commitment therapist's guide

https://lichess.org/@/tackyshrimp/blog/engaging-healthily-with-chess-an-acceptance-and-commitment...
2•tartoran•1h ago•0 comments