2024-01-04 22:34:03
Artificial intelligence (AI) is reshaping the landscape of our society. Political and economic decision-makers are working to seize the opportunities and meet the challenges that AI brings in the private sector, academia, and at the government level. However, there is a lack of clear consensus among AI experts on the future of this technology. This raises the question: how can we anticipate the evolution and impact of AI?
Research Insights: The Largest Survey of AI Researchers
To answer this question, a large-scale survey was carried out among 2,778 researchers having published in leading AI conferences. This survey, carried out in the fall of 2023, is considered the largest of its kind nowadays.
AI Progress and Predictions: Varied Expectations
The forecasts collected indicate that there are at least 50% chance that AI systems reach several milestones by 2028, such as autonomously building a payment processing site, creating songs indistinguishable from those of popular musicians, and autonomously downloading and adjusting a large language model.
Concerning the performance of machines surpassing humans in all areas, the chances are estimated at 10% by 2027 and to 50% by 2047. This represents an advance of 13 years compared to the predictions of a similar survey carried out a year earlier.
According to this survey once more, there is a 5% chance that AI will cause human extinction.
Automation of occupations: A more distant horizon
The probability that all human professions are fully automatable is estimated to be 10% by 2037 and to Only 50% by 2116. This deadline is significantly further away than that anticipated for other AI milestones.
Uncertainties and concerns: Between optimism and pessimism
Despite general optimism (68.3% of respondents envisage positive outcomes from AI superior to humans), a significant share of researchers (48% of optimists) attribute a chance of at least 5% to extremely harmful outcomes, such as human extinction. More than half of respondents believe “substantial” or “extreme” concern is warranted regarding various AI-related scenarios, such as the spread of misinformation, authoritarian control of populations, and worsening inequality.
Geographic and demographic disparities
Notable differences appear depending on the geographic origins of the respondents. For example, researchers trained in Asia anticipate an earlier arrival of human-level AI compared to their counterparts in Europe, North America, or other regions.
AI security research prioritized
There is broad consensus on the need to further prioritize research aimed at minimizing the potential risks of AI systems.
An uncertain path!
The results of this survey reveal a wide variety of perspectives among AI researchers. While the predictions are uncertain, they highlight the fascinating challenges and opportunities presented by the advancement of AI. The implications for society, the economy and our daily lives are far-reaching, highlighting the importance of navigating this era of technological transformation carefully but hopefully.
Source of the study:
1704425613
#chance #human #extinction #study #researchers
