Research Grants

Landing Pages Grant

Key Aim: Create optimized "landing pages" for three high-value target groups.

Vision

  • Awareness of core AI risks and rationality content among several high-impact groups can reduce existential risk.  Three such high-impact groups are:
    • AI researchers, 
    • People interested in optimal philanthropy, and 
    • Bright students looking for career paths.
  • A rational philanthropist may want information on lives saved per dollar; a smart math type may be engaged by simulation probabilities or the doomsday argument. Each is more likely to read about AI risk if the page is organized around their questions.
  • By building such pages, trying variations, and counting click-throughs, we can: (a) understand our potential allies; and (b) directly engage target groups with AI risk.

Steps for realizing this vision

Step 1. Infrastructure and first drafts. 
Build a first draft landing page for each group, together with alternative headlines or web page variants to test. Choose and install an off-the-shelf A/B testing solution. (A/B testing involves randomly assigning different text, layout, or other web page variations, to show to each IP address and then tracking which variants get the most click-throughs. This common technique allows web site creators to better understand and meet visitors' needs.)

Deliverables:
 An initial set of landing pages. 
 A functioning A/B testing solution.
Target date: six weeks from start date[1], to account for possible use of outside professionals for web site design.

Step 2:  Online marketing: Google AdWords and Search Engine Optimization.
Perform industry-standard white-hat Search Engine Optimization actions to help users find our content. Publicize the landing pages by linking to them from relevant websites and by experimenting with Google adwords and with web page variants (using the A/B testing solution installed in step 1.) Using standard web analytics, gather data on which web page variants and which adwords generate the most engagement (as indicated by click-throughs); revise web site as appropriate.

Deliverables:
A more effective set of landing pages, together with data on the pages' effectiveness. 
 Functioning Google Adwords search terms.
Target date: ten weeks from start date. (The lag is useful for gathering user data.)

Step 3:  Audience feedback, and further refinement. Ask members of each target audience to explain what they think when they read the landing page. Generate further website variants in light of target group feedback. Test the variants, and revise the web page as necessary.

Deliverable: A still more effective set of landing pages, together with still more variants and data.
 Target date: fourteen weeks from start date. (Again, the lag is useful for gathering user data.)

Step 4:  Assessment. Assess the cost-effectiveness of this work, as described below.

Deliverable: A cost-benefit analysis of the work done under this grant and of possible extensions, to be posted on the SIAI webpage.
 Target date: six months from start date. (The delay is, again, to allow more user traffic data to accumulate.)

[1] The "starting date" is the date (guaranteed to be within six months of the receipt of grant money) when we have skilled people to allocate to the project.  Extra donations increase our base of skilled people and thereby increase the number of projects we can get to; the lagged start date allows us to find new people, bring them here, and train them.

Total budget: $9,300, consisting of $4,800 of SIAI staff time and $4,500 of professional designer time.

How cost estimates are calculated:

  • Writing web page text and variants, in three iterated steps: 270 hours (30 hours/page * 3 pages * 3 iterations).
  • Querying and listening to target audience reports: 24 hours (1 hour/person * 8 people / target group * 3 target groups)
  • Website programming, setting up analytics, and gathering data: 30 hours
  • Data analysis: 30 hours
  • Graphical webpage design and set-up: 60 hours of professional designer time (7 hours / version * 3 versions/page * 3 pages)
The time can be divided between SIAI staff time and Visiting Fellow time as well as professional designer time for the design of the web pages:
  • SIAI time: 2 person-months (Rounded down slightly, at $2400 / person-month.[1])
  • Professional designer time: 60 hours (at $75 / hour)

[1] This billing rate reflects an estimate of financial outlays for SIAI to create the equivalent of one full-time skilled person-month, including stipend or hosting expenses, workspace, and administrative or management time, and other supporting expenses. Actual person-months may be greater or lower depending on the labor mix for a particular project, with shortfalls made up from general funds. This rate is not reflective of the money Visiting Fellows or interns could earn in the competitive labor market. Think of this as a matched donation. You donate the living expenses; our Visiting Fellows donate the surplus value of their labor.

How this grant, if funded, can be expected to reduce existential risk:

Benefit 1: More people will have the concepts, questions, and background knowledge needed to think well about AI risk.  They will thus be more likely to generate good ideas, to form communities capable of action, and to actually take action to reduce existential risk.

Benefit 2: These pages would teach us (through traffic data) how to better inform and influence audiences through future such pages, and through academic papers.
How these benefits will be assessed:

A brief assessment of the grant's effects will be written up afterward, discussing whether similar grants would be a good use of future funds. The assessment will include:

  • The actual time and money costs of the project;
  • How many visitors each page has received;
  • Whether individuals engaged with these issues report referring potential allies to the pages;
  • Whether anyone who does useful research, donates money, or otherwise reduces existential risk finds these issues through these pages (and tells us);
  • How significant the gains from A/B testing were; how much click-through rates changed as we tried variations;
  • Google hits for these pages; comparison between hits and traffic for these pages, and hits and traffic for other pages about AI risk.


Credit card transactions are securely processed through PayPal.

PayPal – Grant Donation

(200 characters max)