Monkey Score Pro
How much do you think for yourself, and how much do you leave to the AI? Pro measures it, clearly and transparently.
The browser extension for ChatGPT, Gemini and Claude shows your share of the thinking with statistical measures instead of game elements. Everything is analysed on your device.
What Pro measures: 23 metrics in four areas
Every interaction with the AI is captured along reading behaviour, response and task setting, and analysed statistically.
- 1. Reading behaviour
- Reading speed per section (words per minute)
- Visibility time per section (viewport time)
- Share of each answer read
i
How much of an answer was really read. Each section counts by reading speed: up to 350 words per minute fully, up to 800 at 40 %, faster not at all. Example: of 400 words, 200 are read and 100 skimmed: (200 + 40) / 400 = 60 %. From 70 % the answer counts as read, from 30 % as skimmed. - Rating: read, skimmed, scrolled past
- Reading depth for long and short answers
i
Compares how often short answers (under 400 words) and long answers (from 400 words) are read carefully. Example: short 60 %, long 20 %. Long answers often contain the key details but are the most likely to be skimmed. Only the word count is stored, never the text. - Rereading earlier answers
- 2. Response
- Questioning and pushing back
- Own contribution
- Wanting to understand
- Asking for instructions
- Waving through and handing over the task
- Copying without reading, pasting instead of writing
- 3. Task setting
- Goal stated
- Context given
- Limits set
- Format specified
- Length of the prompt
- 4. Statistics
- Share of thinking with 95 % confidence interval
- Trend with significance test
- Median, interquartile range, standard deviation
- Robustness of the data basis
- Trend over 7 or 30 days
- Comparison by AI (ChatGPT, Gemini, Claude)
This produces four core metrics, read carefully, questioned, own contribution, waved through, and up to three concrete insights, for example: “With long answers you read carefully much less often.”
Scientifically sound, not sugar-coated
- Reading is measured via viewport time per section, an established method in reading research.
- Responses are classified by documented rules, not a black box.
- Limits are stated: Pro measures behaviour, not competence, and has not been externally validated.
- Background: studies show that critical thinking declines the more people trust AI. See the studies
PDF report
One click creates a report with all key figures, statistical measures, method and references. Suitable as personal evidence of AI literacy, for example under Article 4 of the EU AI Act. More on AI literacy
Privacy
Everything is analysed locally. Only numbers are stored, never text. No account, no transmission, no ads. Data older than 90 days is deleted automatically.
Pricing
Pro Personal
€4.99one-time
For adults who want to understand and improve how they work with AI.
- All 23 metrics and core figures
- Statistics with confidence interval and significance test
- Insights and PDF report
- Everything stays on your device
- No team report
Mac and iPad (Safari): now on the App Store, €4.99 one-time, one purchase covers both devices. Chrome and Edge: free for now in the Chrome Web Store. Firefox: free for now on Firefox Add-ons.
Pro Business
€3 to €5per active person per month
For companies that want to make AI literacy visible across teams.
- Everything in Pro Personal for all employees, at no extra cost
- Dashboard with team report, trend and comparison by AI
- Anonymous, only from 5 people, no individual scores for managers
- Central rollout by IT, any browser
- Documents for works councils and data protection
In development. Starts with a paid 3-month pilot.
Request a pilotPersonal prices include VAT. Business prices exclude VAT. If you already bought Pro, free months are credited in the business plan.