[2/3] Future of Work with AI Agents: Auditing Automation and Augmentation Potential across the U.S. Workforce (AI column)
Continue. story About the study, I will share the results found.
Positive attitude towards routine automation Contrary to common fears, a significant proportion of workers want to hand over routine and low-value tasks to AI. According to the survey, 46,1Respondents gave a positive assessment of the possibility of automation (higher-net 5-point scale). And before answering, people were asked to think about the risk of losing their jobs and whether they like the task at hand – but even so, almost half of the tasks are desirable for automation. n Free up time for more important work (noted ~69Percentage of automation supporters)
- Get rid of the routine. (~47%) Improve the quality of the result with AI (~46%)
- Avoid stress. (~25%)
Fears and areas of rejection However, the survey confirmed the presence of serious concerns of workers about AI. The most common Distrust in the quality of AI solutions (mentioned ~45% - doubts about the accuracy, reliability of the algorithms)
- Fear of job loss (23%) Lack of human approach (16%) These fears are particularly strong in the creative and humanitarian realms, such as the Art, Design, Media sector. 17% of tasks received a positive evaluation of automation
Four zones of correspondence of desires and opportunities Comparing the desire of employees with the assessments of experts, the authors divided all tasks into 4 category
- “Green light” are tasks that humans are willing to give to AI and for which there are already technical possibilities. Such tasks are the first candidates for the implementation of AI, promising the greatest gains in productivity.
- Red lights are tasks that AI can do, but humans don’t want to automate. Here, the potential implementation of AI can meet resistance or have negative social consequences, so caution is required.
- The R&D zone is a task that employees would love to automate, but current AI models can’t. These areas are promising targets for further AI development in order to meet the explicit demand of users.
- “Low priority” are tasks with low desire for automation and low feasibility of AI; they can be neglected in the near term.
The analysis showed significant inconsistencies: 41The percentage of tasks are either in the red zone or in the “low priority”, that is, a significant part of the current efforts to implement AI is either directed not where people want it, or trying to automate what is not yet possible technology. For example, it turned out that startups from the Y Combinator accelerator often target tasks from the “red” or low-priority zones, whereas many areas are desirable for people. (green zone and R&D) They remain underinvested. This finding highlights the gap between the interests of developers/investors and the needs of employees, and also points to where to redirect efforts to the tasks of the Green Light zone and R&D, where demand is high and technology is lacking.
Preferring cooperation rather than total replacement Most workers prefer not full automation, but partnership with AI. The scale of human participation (HAS) The most common ideal is H3. (partnership)on average 45,2The percentage of respondents would like to work in tandem with an AI agent as an equal partner. More ~35,6% prefer the H4 model, i.e. AI as a human-controlled assistant (People make key decisions). Thus, collectively ~80Percentage of respondents are in favor of maintaining a significant human role in the implementation of AI. Only a small percentage of employees (near 19%) Ready for almost full AI autonomy (H1-H2 levels) Or, conversely, to a very minimal use. (H5). This shows a clear reluctance to outsource work to machines: people want control and critical decisions to be left to them.
Ending in next post.
#AI #Economics #ML #Work