Facilitating Learning Part 3: AI Will Be The End of White Collar Jobs 《促进学习》第 3 部分：人工智能将是白领工作的终结者
While working in industry, we used scripted responses from a corporate database and relied heavily on computer automation when doing our jobs. At this time in my career, I was working in data analytics, and I would receive requests for RPA’s (Rate Plan Analysis) from some of the world’s largest companies. This task would be unimaginable for a human to do on their own as many RPA’s required us to review usage reports that were over 16,000 pages long for the current and previous two months. We had to identify trends and anomalies and make recommendations that would result in measurable cost savings for the client. We had to do this in under 48 hours after receiving the initial request.
在工业领域工作时，我们使用公司数据库中的脚本回复，在工作中严重依赖计算机自动化。在我职业生涯的这个时期，我从事的是数据分析工作，我会收到一些全球最大公司的 RPA（费率计划分析）请求。这项任务对人类来说是难以想象的，因为许多 RPA 要求我们查看当前和前两个月超过 16000 页的使用报告。我们必须识别趋势和异常情况，并提出建议，为客户节省可衡量的成本。我们必须在收到最初请求后的 48 小时内完成这项工作。
Data analytics was my first professionally qualified job after graduating from university in 2010. At the time, I had just graduated from my second university program. Even with my familiarity with computers and my two degrees, I still had to undergo an intensive 3-month training program, and not everyone who was hired made the cut by the end. Once ready to start working on my own, I was able to complete about one large RPA per standard workday. Now, to provide a breakdown of what this process looked like, I would:
数据分析是我 2010 年大学毕业后的第一份专业工作。当时，我刚从第二个大学专业毕业。即使我熟悉计算机并拥有两个学位，我仍然必须接受为期 3 个月的密集培训，而且并不是每个被录用的人最后都能入选。准备好开始独立工作后，我能够在每个标准工作日完成大约一个大型 RPA。现在，我想详细介绍一下这个过程：
- Pull the historical data from the clients’ accounts (e.g., download a PDF copy of the client’s usage reports for the past three months). Furthermore, this would take about 5 minutes to do.
从客户账户中提取历史数据（例如，下载客户过去三个月使用报告的 PDF 副本）。此外，这大约需要 5 分钟的时间。 - Using a custom-made VBA Script (Microsoft Visual Basics), the computer would consolidate all the information into a single Microsoft Excel file (NOTE: In 2010, MS Excel had a theoretical limit of 1,048,576 rows by 16,384 columns, and there were times we would exceed this limit — that is how much data there was on some of these accounts).
使用定制的 VBA 脚本（Microsoft Visual Basics），计算机会将所有信息合并到一个 Microsoft Excel 文件中（注：2010 年，MS Excel 的理论限制为 1,048,576 行 x 16,384 列，有时我们会超过这个限制，这就是其中一些账户的数据量）。 - As the information that we were querying was purely numerical in nature, the computer was able to speed through this task, typically in under 90 seconds. With all the data consolidated into a single spreadsheet, the analyst (that was me) would have to go through the data to identify trends or anomalies within the data.
由于我们查询的信息纯粹是数字性质的，因此计算机能够快速完成这项任务，通常不超过 90 秒。将所有数据合并到一个电子表格中后，分析师（也就是我）需要通过数据来识别数据中的趋势或异常。 - If an anomaly was found, we would have to make note of it and make informed recommendations for the client to act on. For instance, some corporate end users would incur significant overages. While some of these instances were legitimate work expenses, others were not. Let me give you some examples of each; there were times when a user would abuse their corporate device and rack up significant fees for personal downloads such as ringtones. These anomalies would be identified, and the information passed on to the account managers so that they could deduct wages from the employees’ pay cheques. However, in other situations, an end user might incur significant roaming fees due to added responsibility to their Job Description. In this case, changes to the user’s corporate plan to reduce overages would be recommended to the client, and this whole process would take a well-trained analyst about five to seven hours to complete.
如果发现异常，我们就必须将其记录下来，并提出明智的建议供客户采取行动。例如，一些企业最终用户会产生大量超额费用。其中有些是合法的工作支出，有些则不是。让我分别举几个例子：有时，用户会滥用公司设备，为铃声等个人下载支付大量费用。这些异常情况会被发现，并将信息传递给客户经理，以便他们从员工的工资支票中扣除工资。然而，在其他情况下，最终用户可能会因工作职责增加而产生大量漫游费用。在这种情况下，将向客户建议修改用户的公司计划，以减少超额费用，整个过程需要一名训练有素的分析师花费大约五到七个小时才能完成。 - Finally, the last step was to format the data and write a short and concise response to the client. This may seem simple enough, but we had stringent guidelines on how to do this. For instance, the formatting of each cell, header, border, etc., had to match the company’s corporate colours. Moreover, finally, the height and width of columns had to be set to exact measurements. And the response to the client had to be written in a precise way.
最后一步是对数据进行格式化，并写出简明扼要的回复给客户。这看似简单，但我们有严格的指导原则。例如，每个单元格的格式、标题、边框等都必须与公司的企业颜色相匹配。此外，栏目的高度和宽度也必须精确定位。给客户的回复也必须写得准确无误。
In essence, we were trained to be like computers: decisive, analytical, and void of human emotion. Part of this was to avoid legal liability, and the other reason was to ensure consistency in the presentations of our reports for our clients. For instance, we could never write, “I am sorry to hear that your account went over balance this month,” as the word sorry implies fault. This one statement could then be used as grounds to initiate a lawsuit for a bad RPA recommendation. Therefore, displays of empathy and compassion were not allowed. Every response was carefully scripted by the legal department and added to a corporate database, and our responses to the client were based on these comment banks. We essentially trained humans to become the perfect computer!
从本质上讲，我们被训练得像电脑一样：果断、善于分析、没有人情味。这样做的部分原因是为了避免承担法律责任，另一个原因是为了确保向客户提交报告时的一致性。例如，我们决不能写：”很遗憾听到您的账户本月余额超支”，因为 “遗憾 “一词意味着过错。这一句话可能会被用作对不良 RPA 建议提起诉讼的理由。因此，不允许表现出同情和怜悯。每条回复都由法律部门精心编写，并添加到公司数据库中，我们对客户的回复都是基于这些评论库。我们从本质上将人类训练成了完美的计算机！
Now you might ask why this is important. Well, now, with the power of Chat GTP version 4.0, my old data analytics job, a job I had to fight for even with two university degrees… well, that job won’t exist for long. Based on the evolution of AI technologies, I predict a computer could do the entire RPA in under 5 minutes, and as computer processing power increases, that time will only decrease even further. All that will be needed is for a user to initiate a query such as “download and consolidate the last three months of user data, analyze and identify user anomalies and trends, make recommendations to minimize overages, and format the report using the corporate colours of company X.” It will only take an instant, and the computer will be able to generate the entire report on its own. How will humans ever be able to compete with that?
现在你可能会问，这有什么重要的？现在，有了 Chat GTP 4.0 版的强大功能，我以前的数据分析工作–一份即使拥有两个大学学位也要为之奋斗的工作……但是这份工作未来可能不会存在太久了。根据人工智能技术的发展，我预测计算机可以在 5 分钟内完成整个 RPA 操作，而且随着计算机处理能力的提高，这个时间还会进一步缩短。届时，用户只需发起一个查询，如 “下载并整合过去三个月的用户数据，分析并识别用户异常和趋势，提出建议以尽量减少超额使用，并使用 X 公司的企业色来格式化报告”。只需一瞬间，计算机就能自行生成整个报告。人类如何能与之匹敌？
Regrettably, that is not the only type of job AI will affect in the coming years. Professional careers like accountants who track numerical expenses and data analysts looking for trends and anomalies in data can easily be replaced by a computer better suited to analyzing and tabulating numerical data. Computers can crunch numbers faster and more accurately than humans, and they do not get tired of doing repetitive tasks and do not make careless mistakes. AI can also easily replace lawyers as most legal work analyzes evidence dockets against legal precedence. And ironically enough, programmers and coders are also at risk as most coding is just reusing lines of code which are often stored in a database. Regrettably, to our dismay, most school systems have been heavily promoting coding over the past decade at the expense of all other STEM fields of study. What do all these jobs have in common though? Put simply, they are all high-paying white-collar desk jobs that are the crowning jewel of our current education system. A system of education that is heavily skewed towards summative assessments, standardized tests, analytical thinking and less concerned with process-driven inquiry-based learning.
遗憾的是，这并不是未来几年人工智能将影响的唯一工作类型。追踪数字支出的会计师和寻找数据趋势和异常的数据分析师等专业职业，很容易就会被更适合分析和制表数字数据的计算机所取代。与人类相比，计算机可以更快、更准确地计算数字，而且不会厌倦重复性工作，也不会犯粗心大意的错误。人工智能还可以轻松取代律师，因为大多数法律工作都是根据法律先例分析证据卷宗。具有讽刺意味的是，程序员和编码员也面临风险，因为大多数编码工作都是重复使用代码行，而这些代码行通常都存储在数据库中。遗憾并令人感到沮丧的是，在过去十年中，大多数学校系统都在大力推广编码技术，而忽略了所有其他 STEM 领域的学习。这些工作有什么共同点呢？简单地说，它们都是高薪的白领文职工作，是我们当前教育体系的皇冠。这种教育体系严重偏向于终结性评估、标准化考试和分析性思维，而较少关注过程驱动的探究式学习。
So, what does inquiry and project-based learning look like?
那么，探究式学习和基于项目的学习是什么样的呢？
Well, it looks like experiential learning and STEM education! A framework that focuses on critical thinking and less on data analytics. A system of learning that capitalizes on what we as humans are uniquely qualified for. Something that we excel at. So, you might ask, why are most school systems so heavily skewed toward analytical thinking and processes? Why are we trying to make humans try to compete with computers on their turf? We have stacked the deck against ourselves and are trying to compete in an area in which we will never be able to compete. Nevertheless, that does not need to be the case. Computers are great at analytical thinking, whereas humans are well suited to critical thinking, which we have been saying for years. Regrettably, the lure of summative assessments and standardized tests that focus on knowledge recall and analytical processes is hard to abandon as they are far easier to evaluate than wholistic, process-driven education that focuses on inquiry and the pursuit of discovery.
这看起来像是体验式学习和 STEM 教育！一个注重批判性思维而非数据分析的框架。一个学习系统，利用我们人类独有的资质、我们擅长的东西。那么，你可能会问，为什么大多数学校系统如此偏重分析性思维和过程？为什么我们要让人类在计算机的地盘上与计算机竞争？我们对自己不利，试图在一个我们永远无法竞争的领域进行竞争。然而，事实并非如此。计算机擅长分析性思维，而人类更适合批判性思维，这一点我们已经说了很多年。遗憾的是，终结性评估和标准化考试注重知识记忆和分析过程，很难放弃这种诱惑，因为它们远比注重探究和追求发现的全面、过程驱动型教育更容易评估。