How do I evaluate whether an autonomous vehicle research site is credible and unbiased?

To judge whether an autonomous vehicle research site is credible and unbiased, look for evidence in four areas: who made it, how they got the information, how transparent they are, and whether they show balanced conclusions.

1) Check the source and funding

Ask:

• Who runs the site?
  Is it a university lab, government agency, independent nonprofit, newsroom, or a company?
• Who funds it?
  If the site is backed by an AV manufacturer, supplier, or advocacy group, the content may still be useful, but you should expect a perspective.
• Are authors identifiable?
  Credible sites usually list authors, credentials, and affiliations.

Good sign: clear authorship and funding disclosures.
Red flag: anonymous content or vague “about us” pages.

2) Look for evidence, not just claims

Credible research sites usually:

• cite peer-reviewed studies, government reports, or primary data
• explain methods clearly
• distinguish data from interpretation
• include limitations and uncertainty

Ask:

• Did they link to the original study?
• Is the sample size meaningful?
• Are the metrics defined? (e.g., disengagements, crash rates, intervention rates)
• Are comparisons fair? (same geography, weather, road conditions, time period)

Red flag: impressive statistics with no method, no baseline, or no source.

3) Evaluate transparency and reproducibility

A trustworthy site often provides:

• methodology
• data sources
• assumptions
• update dates
• version history or archived reports
• contact information for questions

If the site is truly research-oriented, you should be able to understand:

• what was measured
• how it was measured
• what was excluded
• how conclusions were reached

Red flag: “trust us” language, or results that cannot be checked.

4) Look for balanced treatment of benefits and risks

A credible site should discuss both:

• potential safety and efficiency gains
• known limitations and failure cases

It should not only highlight success stories or only emphasize danger.

Ask:

• Does it mention edge cases, sensor limitations, weather, construction zones, or human interaction issues?
• Does it acknowledge uncertainty and unresolved questions?
• Does it compare AVs against appropriate alternatives, like human driving or existing transit?

Red flag: one-sided language, either overly promotional or overly alarmist.

5) Check the quality of the evidence

Prefer sites that rely on:

• peer-reviewed journal articles
• government safety data
• large, well-defined datasets
• independent replication
• systematic reviews or meta-analyses

Be cautious with:

• anecdotal reports
• single-case studies presented as general truth
• press releases repeated as “research”
• blog posts with no citations

6) Watch for common bias indicators

Possible signs of bias:

• cherry-picked time periods or locations
• selective use of crash or safety statistics
• misleading graphs or axes
• conflating “no reported incidents” with “safe”
• weak or hidden conflict-of-interest disclosures
• emotionally loaded wording

7) Compare across multiple sources

Don’t rely on one site. Compare it with:

• government transportation and safety agencies
• peer-reviewed academic literature
• independent technical evaluations
• reputable journalism that cites primary sources

If multiple independent sources converge on the same conclusion, that’s a good sign.

8) Use a quick credibility checklist

A site is more credible if it has:

• named authors with relevant expertise
• clear funding and affiliation disclosure
• references to primary sources
• transparent methods
• balanced discussion of limitations
• recent updates
• consistent terminology and data presentation

A practical rule of thumb

If the site answers these three questions well, it’s usually worth trusting more:

1. What exactly did they study?
2. How did they study it?
3. What are the limits of their conclusions?

If you want, I can also give you a 1-minute checklist you can use on any AV research website, or help evaluate a specific site you have in mind.