https://arxiv.org/html/2401.02843v3

In both 2022 and 2023, respondents gave a wide range of predictions for how soon HLMI will be feasible (Figure 3 ). The aggregate 2023 forecast predicted a 50% chance of HLMI by 2047, down thirteen years from 2060 in the 2022 survey. For comparison, in the six years between the 2016 and 2022 surveys, the expected date moved only one year earlier, from 2061 to 2060 1 1 1

The aggregate 2023 forecast predicted a 10% chance of HLMI by 2027, down two years from 2029 in the 2022 survey. See Appendix B for a table comparing this with Full Automation of Labor, which is discussed below.

The aggregate 2023 forecast predicted a 50% chance of FAOL by 2116, down 48 years from 2164 in the 2022 survey (Figure 3 ). We checked if this difference was significant for participants who received the question in the fixed-probabilities framing, and found that it was (p = .0052, Yuen’s test (bootstrap version); see Appendix F ). There is about a 70-year difference between the mean 50% prediction for HLMI and the mean 50% prediction for FAOL. We discuss this surprising finding in the next section, “Framing effect of HLMI vs FAOL.”

Predictions for a 50% chance of the arrival of FAOL are consistently more than sixty years later than those for a 50% chance of the arrival of HLMI. This was seen in the results from the surveys of 2023, 2022, and 2016. This is surprising because HLMI and FAOL are quite similar: FAOL asks about the automation of all occupations; HLMI asks about the feasible automation of all tasks. Since occupations might naturally be understood either as complex tasks, composed of tasks, or closely connected with one of these, achieving HLMI seems to either imply having already achieved FAOL, or suggest being close.

The previously-observed framing effect was again observed in this survey. For example, the year with a 50% chance of HLMI from participants answering in the fixed-year frame (34 years) was twice as far into the future as that for participants answering in the fixed-probability frame (17 years). However, it’s notable that even the larger of these two is shorter than 2022’s combined forecast (37 years), demonstrating a substantial shift of predictions closer to the present (Fig 18 in Appendix B ).

While the range of views on how long it will take for milestones to be feasible can be broad, this year’s survey saw a general shift towards earlier expectations. Over the fourteen months since the last survey [ Grace et al., 2022 ] , a similar participant pool expected human-level performance 13 to 48 years sooner on average (depending on how the question was phrased), and 21 out of 32 shorter term milestones are now expected earlier.

There are signs in this research and past surveys that these experts are not accurate forecasters across the range of questions we ask. For one thing, on many questions different respondents give very different answers, which limits the number of them who can be close to the truth. Nonetheless, in other contexts, averages from a large set of noisy predictions can still be relatively accurate [ Surowiecki, 2004 ] , so a question remains as to how informative these aggregate forecasts are.