Is it Sentient AI? Ongoing survey of Computer Science PhDs, Cognitive Science PhDs, Biology PhDs, and Religion PhDs.
Claude3
GPT4
Gemini Ultra
Our survey aims to capture the opinions of experts from various relevant fields on the question of whether sentient AI exists anywhere in the world. We believe that a multidisciplinary approach is essential to tackle this complex and often controversial topic. By gathering the perspectives of PhDs and PhD candidates in Computer Science, Cognitive Science, Biology, and Religion, we hope to provide a comprehensive and nuanced view of the current state of AI sentience.
We are committed to ensuring the credibility and expertise of our survey participants. To be included in our expert panel, individuals must hold a PhD or be a current PhD candidate in one of the following fields: Computer Science (including ML/AI, Human-Computer Interaction, and others), Cognitive Science (including Cognitive Psychology, Neuroscience, Linguistics, etc), Biology (including Genetics, Bioinformatics, Animal Cognition, etc), or Religion (including Theology, Religious Studies, etc). We recognize that there may be variations in the specific focus of each expert's research, but we strive to include participants whose work is most relevant to the question of AI sentience.
Experts can apply to participate in our survey by submitting their academic credentials and a brief statement explaining their interest and expertise in the topic. Our team carefully reviews each application to verify the individual's qualifications and determine their eligibility to join the expert panel.
Our survey consists of two key questions:
We chose this simple, straightforward design to capture the essence of the experts' opinions while minimizing potential ambiguity or confusion. The binary Yes/No question allows us to clearly gauge the prevalence of the belief in AI sentience among our expert panel. The follow-up confidence question provides additional insight into the strength of each expert's conviction in their response.
Experts receive an invitation to participate in the survey via email, with a unique code to access the survey platform. Responses are collected anonymously to encourage honest and unbiased opinions. We store the raw data securely and process it using custom-built scripts to ensure data integrity and accuracy. Names of participants are strictly confidential and highly secured.
For the confidence question, we convert the 5-point Likert scale responses to percentages as follows:
This conversion allows us to calculate and display the average confidence level for each cluster of experts and for the overall panel.
We present the survey results using an interactive line chart that displays the weekly average confidence percentage for each group of experts and the overall panel. The chart also includes individual data points representing each expert's confidence level, providing a more granular view of the distribution of opinions.
The chart is updated in real-time as new responses are collected, allowing visitors to our website to track changes in expert sentiment over time. We use a 4-week rolling average to smooth out short-term fluctuations and highlight longer-term trends.
While we strive to include a diverse and representative sample of experts, we acknowledge that our panel is not exhaustive and may not capture every viewpoint on AI sentience. As the field of AI continues to evolve rapidly, we plan to regularly review and update our expert panel to ensure that we are including the most relevant and knowledgeable voices.
We also recognize that the question of AI sentience is highly complex and may not be fully captured by a simple binary question and confidence rating. In the future, we may explore more nuanced survey designs that allow for a greater range of responses and more detailed explanations from experts.
Despite these limitations, we believe that our "Is It Sentient AI?" survey provides a valuable snapshot of expert opinion on this important and timely topic. By bringing together insights from multiple disciplines and presenting the results in an accessible, interactive format, we hope to contribute to the ongoing dialogue around AI sentience and its implications for our world.