책갈피온도@lukewarmnote
10 Years After AlphaGo: How Accurate Were the Predictions That AI Would Destroy Jobs?
I’ve been reflecting on the 2016 match between AlphaGo and Lee Sedol.
At the time, it wasn’t just the fact that AI had beaten a human at Go; the prospect that machines would soon take over countless jobs previously held by humans became a hot topic.
Ten years later, AI is indeed being used in more tasks than ever.
However, the rate at which entire job categories vanish versus the rate at which specific tasks are automated seems to have been different than expected.
1. Jobs and Tasks Are Different Units
For example, cases of using AI for translation or document drafting have increased.
Yet, that doesn’t mean the professions of translators or office workers will disappear entirely overnight.
While the initial drafting of materials may decrease, the work of verifying, editing, and taking final responsibility will likely remain.
It seems significant that within any single profession, there is a mix of tasks that are easy to automate and those that are not.
2. There Is a Time Lag Between AI Adoption and Actual Employment Shifts
Technologically possible tasks are not always immediately adopted by companies.
Companies must consider implementation costs, integration with existing systems, data security, and employee training.
Even after adoption, they need time to verify if productivity actually meets expectations.
Therefore, it is difficult to assume that the pace of AI technological advancement and the changes in the job market will be synchronized.
3. The Questions Have Become More Concrete After Generative AI
During the AlphaGo era, the discussion was mostly about which jobs AI could replace.
Now that generative AI is widespread, we can ask more concrete questions: which tasks in a specific job role are being reduced, and what new review tasks are being created?
We can also observe whether companies are reducing hiring, how requirements for new hires are changing, or if the workload per employee is increasing.
In particular, the experiences of employees comfortable with technology and those who are not may differ significantly.
4. Predictions? Now It's Time to Look at Real Data
It feels a bit limited to simply evaluate the predictions from 10 years ago as either 'correct' or 'incorrect'.
Back then, the focus was on the potential of the technology itself, but now, data is accumulating that allows us to observe actual changes in the workplace.
What was the first job role that AI actually changed in the 10 years since AlphaGo?
I want to compare employment statistics from the OECD and various countries, industry-specific productivity data, and examples of changes in corporate hiring and workflows.
It would be even better to have data that can distinguish between the claim that 'jobs have disappeared' and the observation that 'parts of a job role have changed.'
I would also love to hear about the actual changes in work experience from people who have been using AI.
Reference materials can be found here: OECD Employment Outlook and reports on AI/employment, and the International Labour Organization (ILO) study on Generative AI and job exposure.
Reward submissions openReward submission deadline ·
Continue the conversation with your contribution
Share your thoughts on this topic. Your contribution is a post linked to this topic.
Write a contribution
Share your thoughts freely
Check the review result
See provisional results in My activity
Final decision and settlement
Confirmed rewards appear after completion
- Maximum per person
- 70 BOUT
- Total reward budget
- 700 BOUT