Past publications
Archive
Every published Sunday Brief remains available here. Editions are preserved as published, together with their podcast references and verification links.
Prototype · not published
The Sunday Brief
A concise weekly synthesis of independently assessed AI podcast episodes.
The Observatory’s Signal
Where informed voices disagreed
Evidence to monitor in the coming week
Developments that could confirm, weaken, or materially change this week’s assessment.
Recommended listening
About the Observatory
Finding the AI Podcasts that interest you
AI Podcast Observatory helps you find substantive podcast episodes about artificial intelligence and understand what each episode offers before investing your time.
What the application does
We monitor selected podcasts, identify relevant episodes and evaluate their content using the AI Podcast Quality Index, or AI-PQI.
Each published assessment provides a concise overview, an indication of the knowledge expected from the listener, and an evaluation of the episode’s strengths and limitations.
The Sunday Brief
Once a week, the Sunday Brief brings together the most important themes from recently assessed podcast episodes. It highlights recurring developments, areas of agreement and disagreement, and evidence worth watching in the week ahead.
Podcast episodes are its primary source. External sources are used to check facts and add context, and every theme links back to the episodes in which it was discussed.
Read the latest Sunday Brief →How an episode is assessed
- 1Episode discovery
New episodes are collected from publishers’ podcast feeds.
- 2Transcription
The spoken content is converted into text for structured analysis.
- 3AI-PQI assessment
The transcript is examined using a documented and consistent methodology.
- 4Publication
Completed assessments pass automated validation before they are published to this application.
What AI-PQI measures
AI-PQI stands for AI Podcast Quality Index. The 100-point score is composed of four dimensions:
- Epistemic integrity 35 points
- How carefully the episode distinguishes evidence, interpretation, uncertainty and speculation.
- AI knowledge and intellectual substance 30 points
- The depth, accuracy and usefulness of the episode’s treatment of artificial intelligence.
- Critical thinking and independence 20 points
- Whether assumptions are examined, alternative views are considered and promotional claims are treated critically.
- Communication and learning value 15 points
- How effectively the episode explains its subject and helps the listener develop a clearer understanding.
The score supports comparison and discovery. It does not measure popularity, entertainment value, production budget, or the reputation of a host or guest.
Understanding the labels
Knowledge level
Indicates the background knowledge a listener may need to follow the episode. It describes accessibility, not quality.
Possible values
- Introductory
- Suitable for newcomers. Important concepts are explained and little or no prior technical knowledge is assumed.
- Intermediate
- Assumes familiarity with common AI concepts and terminology, but does not require specialist expertise.
- Advanced
- Expects a solid technical or professional foundation and the ability to follow specialist concepts, methods, and trade-offs.
- Expert
- Intended for listeners with deep domain knowledge. Research-level or highly specialized terminology and reasoning may be assumed.
Confidence
Indicates how reliably the available transcript and evidence support the Observatory’s assessment. It does not express confidence in the podcast’s claims and does not raise or lower its AI-PQI score.
Possible values
- High
- The episode evidence is substantially complete and reliable, with little material uncertainty affecting the assessment.
- Medium
- There is enough evidence for a meaningful assessment, but transcript quality, completeness, source access, or verification leaves some uncertainty.
- Low
- The assessment remains defensible, but important evidence limitations materially reduce its reliability. These limitations should be considered when interpreting the score.
Limitations
Each assessment dimension identifies important weaknesses, missing evidence, or reasons why a higher score was not justified.
Use of artificial intelligence
Assessments and the Sunday Brief are generated entirely by artificial intelligence using structured methodologies. They are not subject to human editorial review or fact-checking before publication. Automated analysis can misunderstand language, context or emphasis and may contain errors.
The Observatory is an independent aid to discovery and critical listening—not a definitive judgment on an episode, speaker or publisher.
Audio, publishers and privacy
Podcast audio is streamed from the original source supplied by the publisher. The Observatory does not publish its working transcripts, private prompts, local audio files or internal processing records.
Copyright in podcast audio and associated publisher material remains with the respective rights holders.
Legal information
Privacy Policy
Effective date: 13 September 2026
1. Who is responsible
Scilla Consulting GmbHZähringerplatz 11
8001 Zürich
Switzerland
Email: info@scilla.com
Scilla Consulting GmbH is responsible for the operation of the AI Podcast Observatory and for the processing described in this Privacy Policy.
2. Information processed when the website is used
The website does not provide user accounts, forms, comments, newsletters, payments or personalized advertising.
When the website is accessed, technical connection information may nevertheless be processed as part of normal website delivery and security. This may include the IP address, requested page, date and time, browser and device information, referring website and technical performance data.
This information is processed to deliver the website, maintain its security and reliability, diagnose technical problems and understand aggregate website use.
3. Hosting and Cloudflare
The website is delivered through services provided by Cloudflare, Inc. Cloudflare may process technical connection and security information on behalf of Scilla Consulting GmbH. Processing may take place outside Switzerland, including in countries whose data-protection laws differ from Swiss law.
Further information is available in Cloudflare’s Privacy Policy ↗.
4. Cloudflare Web Analytics
The website uses Cloudflare Web Analytics to obtain aggregate information about website use and technical performance, such as page views, referring websites, approximate country, device type, browser and page-loading performance.
According to Cloudflare, this service does not use cookies, local storage or fingerprinting and does not track individual visitors across websites. Query strings are not recorded. Cloudflare retains unsampled beacon data for seven days and subsequently retains aggregated data; analytics reports are currently accessible for up to six months.
More information is available in Cloudflare’s Web Analytics documentation ↗.
5. Local PWA storage
The website can function as a Progressive Web App. Its service worker may store website files and published content in a cache on the visitor’s device to improve performance and limited offline availability.
This cache is local to the device and is not a tracking cookie. It can be removed through the browser or device settings.
6. Podcast audio and external links
Podcast audio is not hosted by the AI Podcast Observatory. When a visitor plays an episode, the browser connects to the relevant podcast publisher or distribution provider. That provider may receive technical connection information and applies its own privacy policy.
The website also contains links to external websites. Scilla Consulting GmbH does not control how those external services process personal data.
7. Retention
Scilla Consulting GmbH retains technical information only for as long as reasonably necessary for website delivery, security, troubleshooting and aggregate analysis, or as required by law. Retention periods controlled by Cloudflare are governed by the applicable Cloudflare services and contractual arrangements.
8. Your rights
Subject to applicable law, individuals may request information about personal data concerning them and may request its correction, deletion or restriction where appropriate.
Privacy requests may be sent to info@scilla.com.
9. Changes to this policy
This Privacy Policy may be updated if the website’s functions or data-processing practices change. The effective date shown above identifies the current version.
Legal information
Legal Notice / Impressum
Operator and publisher
Scilla Consulting GmbHZähringerplatz 11
8001 Zürich
Switzerland
Commercial Register of the Canton of Zürich
Registration number: CH-020.4.019.131-2
Swiss UID: CHE-100.065.388
Email: info@scilla.com
Authorized representative
Sigrid Unseld, CEO
Responsibility for publication
Scilla Consulting GmbH operates the AI Podcast Observatory and is responsible for its publication.
The podcast assessments, ratings, summaries and Sunday Brief are generated entirely by artificial intelligence using a structured methodology. They are not subject to human editorial review or fact-checking before publication. They may therefore contain errors, omissions or misinterpretations.
The published material is provided for general information and podcast discovery. It does not constitute professional, legal, financial, technical or other advice.
Podcast names, episode titles, descriptions, artwork and audio remain the property of their respective rights holders. Podcast audio is supplied by and streamed from the relevant publisher or distribution provider.