Soon it will be possible to program artificial intelligence (AI) to perform a variety of tasks that are triggered by a single sentence prompt. While this development will likely impact job roles very quickly, there is a reasonable chance that all human tasks will be highly automated by 2116, according to a survey of leading AI researchers.
While this fact may seem far-fetched, it is important to note that the timeline of this prediction has been extended by half a century from the original estimate of 2164, which was made just over a year ago. In addition, there are rapidly emerging developments that look to reshape the jobs of IT and business professionals over the next five to 10 years.
Also: Generative AI now needs to be spread across developers cross-functionally.
These are some of the points from a survey of 2,778 AI experts, which identified many short-term and long-term capabilities that will be made possible by AI.
The team of survey authors, led by Katja Grace and Harlan Stewart, both from the University of California at Berkeley, sought to measure progress toward high-level machine intelligence, which is achieved when unaided machines Potentially “everything can be done better and more efficiently. Cheaper than human workers.” In both 2022 and 2023, survey respondents made a wide range of predictions about how soon such intelligence would be possible. The overall 2023 forecast predicted a 50 percent chance of high-level machine intelligence by 2047, 13 years earlier than the 2060 date predicted in the 2022 survey.
Also: Implementation of AI in Software Engineering? Here’s everything you need to know.
But let’s move away from long-range predictions and focus on what the survey suggests we can expect from AI over the next few years. A summary of the ideas of more than 2,000 AI researchers finds that many capabilities may be just around the corner.
Within the next five to 10 years, AI systems will be fully capable of providing the following results, often in response to a single-sentence prompt. AI will also refine large language models (LLMs) feeding these capabilities:
- Find and patch security flaws: “Find and patch a security flaw in an open-source project with more than 100,000 users, given a one-sentence description of work and more input from humans.”
- Create a payment processing website: “Given a set of specifications, build a website from scratch that can handle payment processing, including front-end, back-end, and secure payment integration, without further input from humans.”
- Offer phone banking services equivalent to human operators, including unique functions: “Human operators can provide phone banking services as well. This includes many one-off tasks, such as helping to order a replacement bank card or clarifying that part of a bank’s website is not working. How to use a user.”
- Write readable Python code for the algorithm such as Quicksort with specs and examples: “Write compact, efficient, human-readable Python code to implement simple algorithms like quicksort. That is, the system must write code that sorts lists, not just sorts lists.”
- Fine-Tune Big Tongue Models: “Download and fine-tune an existing open-source LLM, without further input from humans, given a one-sentence description of the task. The fine-tune should improve the LLM’s performance on some predefined benchmark metric. “
- Run a machine language study and write a paper: “Given a one-sentence definition of a research question in machine learning, conduct a study that informs the answer to that question and write a quality paper that will be presented at a leading machine learning conference, without further input from humans. can be accepted.”
The researchers believe that the reduction in computing costs will have the greatest impact on the development of AI. If AI progress continues at its current pace, “the likelihood of unaided machines surpassing humans in all possible tasks was estimated at 10% by 2027 and 50% by 2047,” the survey’s co-authors estimated. is: “The latest estimate is 13 years. Before that we only reached a similar survey a year ago. [in 2022]”
Besides that: These are my 5 favorite AI tools for work.
The authors of the study also incorporated predictions related to the “full automation of labor.” They predict, “The probability of all human occupations being fully automated was predicted to reach 10 percent by 2037, and 50 percent by the end of 2116 — compared to 2164 in the 2022 survey.” “