Original research
Nobody knows how many Alaska businesses actually use AI. We are going to find out.
Every claim about AI adoption in this state is a national statistic with Alaska pasted on top. This is a real count, by industry, including the pilots that got abandoned and what stopped them. Three minutes, anonymous by default, and we publish the methodology and the raw data alongside the findings.
Written by Kristian Baer, Northtek · Anchorage, Alaska · Updated 2026-08-21
The short answer
What is the Alaska AI Operations Index?
An annual survey measuring how Alaska businesses actually use AI in their operations, run by Northtek. It asks what has been deployed, what was piloted and abandoned, what stopped it, and where an organization’s data is allowed to live - broken out by industry and company size. Existing figures on AI adoption are national studies that say nothing about a state where half the communities have no road access and a large share of the economy runs on resource extraction, federal contracting, and Alaska Native corporations. Responses are anonymous unless you choose otherwise, the methodology and sample size are published with the findings, and the anonymised raw data is released as a CSV so anyone can check the analysis or run their own.
The survey
Four required questions, six optional, about three minutes. Anonymous unless you choose to give us an email at the end.
Recorded. Thank you.
That is one more real data point in a set that does not exist anywhere else yet. If you left an email we will send the findings and the raw data when they publish.
Methodology
The methodology, published before we have a single response
A statistic is only worth citing if you can see how it was produced. This is written now, not after we see whether the numbers are interesting.
Who is asked
Businesses and organizations operating in Alaska, reached through industry association mailing lists and newsletters, direct outreach, and an open link on this page. Distribution channels are listed by name in the published report.
What that means for the result
This is a self-selected convenience sample, not a random one. Organizations already interested in AI are more likely to respond, which almost certainly biases adoption upward. We will state that on the report itself rather than in a footnote, and we will not compute a margin of error, because one calculated on a self-selected sample is decoration.
Minimum before we publish anything
Sixty complete responses overall, and we will not break out any industry with fewer than eight. A percentage of five respondents is not a finding, and reporting it would be the fastest way to lose the credibility this exists to build.
What gets released
The findings, the full question wording, the response counts per question, and the anonymised raw data as a CSV. Free-text answers are released only where they cannot identify the respondent, and quoted only without attribution.
What we will not do
We will not publish a number we did not measure, we will not fill a thin industry segment with an estimate, and if the response count falls short we will say the survey did not reach the bar rather than publishing anyway.
Straight answers
Is this anonymous?+
By default, yes. None of the questions ask who you are. Contact details are optional, stored separately from the answers, and used only to send you the results. If you leave them blank the response still counts.
Is this a sales funnel?+
Not in the sense that matters: you can complete it without giving us any way to contact you, and we will still send nothing. We built it because a firm that publishes the only real numbers on AI adoption in this state becomes the source people cite, and that is worth more to us than a list of emails.
What if we are not using AI at all?+
Then you are one of the most valuable responses in the set. A survey that only hears from adopters produces a number that is wrong in a predictable direction, and "not looking at it" is a legitimate and probably common answer we need counted.
Will you name our company?+
No. Nothing is attributed. Free-text answers are quoted only where they cannot identify the respondent, and industries with fewer than eight responses are not broken out at all, which is as much about protecting anonymity in a small state as it is about statistics.
When do the results come out?+
Once the response count clears the threshold in the methodology above. If it does not clear it, we will say so publicly rather than publishing a number built on too few responses.
Can I see the raw data?+
Yes. The anonymised responses are published as a CSV alongside the findings so you can check the analysis or run your own. A benchmark you cannot reproduce is a marketing figure.
Three minutes, and you get the results first.
Every response makes the number more real. If your organization is not using AI at all, that is still a data point we need.
Anchorage, Alaska · info@northtek.io · (907) 903-4353