使用ChatGPT来帮助将信息处理成熟悉的格式。

新加坡科技周大数据与人工智能世界在本周举行。我注册了参加,但直到最后一刻才抽空过去。自然而然,我想浏览一下节目表,快速了解展览期间发生的各种演讲的主题、内容、地点和时间。看着节目清单(如下截图),我有几个想法。

ChatGPT中文站

我觉得节目的格式不符合我对同时运行多个错位时间的事件所期待的直观格式。虽然我并不是UI或UX专家,但将5个地点并排列出,并且持续的方式给人一种时间相关且连续运行的印象。

不幸的是,情况并非如此,我发现自己不得不上下滚动,努力根据时间匹配,因为没有统一的锚定点来确定时间顺序。试图获得节目的全面概览,以解决同时进行讨论的冲突,找出自己感兴趣的内容,一直让我产生这样的想法。

如果我可以以表格形式呈现这个,那就会容易得多。

自然而然地,将文本复制粘贴到电子表格中是第一想到的方法,但不出所料,复制的文本变得非常难看。

Big Data & AI World Keynote Theatre
Hall F, Basement 2

10:00 10:20
Augment Data Management with AI
Big Data & AI World Keynote Theatre
Sachin Tonk
Sachin Tonk, Deputy Chief Data Officer - GovTech
10:25 10:45
The Impact of Generative AI
Big Data & AI World Keynote Theatre
Introduction to Gen AI What are the potential use cases? Why does it matter to employment and business in Singapore? What are the risks? What is IMDA’s role?
Dr Eric Tan
Dr Eric Tan, Assistant Director (AI Data Engineering) - IMDA
10:50 11:10
Unlocking Growth with Customer Behavioural Data
Big Data & AI World Keynote Theatre
Ben Lu
Ben Lu, Director, Solution Consulting, APJ - Amplitude
11:15 11:35
Democratizing Geospatial Data: From Data Collection to AI + Cloud Processing, and Visualization
Big Data & AI World Keynote Theatre
Key takeaways: How geospatial is collected The issues faced after the data is collected How is the cloud leveraged to store and process the large amounts of geospatial data How is the cloud leveraged ...
Kevin Lu
Kevin Lu, CTO - Skand
11:40 12:00
Innovating at 10X: Building Data & AI solutions with Amro & Domo
Big Data & AI World Keynote Theatre
Overview of Mohammed’s experience at Domo and McKinsey in Building Data Solutions, including common challenges faced, and approaches to increase time to value 10x Deep dive into how AMRO built an econ ...
Mohammed Aaser
Mohammed Aaser, Chief Data Officer - Domo
12:55 13:15
... ...

我曾经简要考虑过爬取网页,将详细信息放入数据框架中,但当时已经很晚,听起来并不吸引人。我几乎要打印出节目列表,以便更好地了解和突出感兴趣的演讲,但突然想到了一个更好的办法。

ChatGPT (聊天GPT)

我回忆起我的朋友和导师通过API展示ChatGPT的一个用例,作为一个摘要生成器或情感分析工具。我想,也许现在是时候尝试一下,看看它在识别未标记但有重复结构信息方面的强大能力。

很遗憾,ChatGPT无法直接评估网页。尽管如此,我决定测试用户界面(而不是通过API,因为这是一次性事件)。

虽然ChatGPT很聪明,但仍然需要明确你在输出中想要什么,并且清楚地提供信息有助于它识别信息的微妙之处。因此,给它一些背景和解释将提供的信息以及所需的输出,我将整个复制的文本块释放到文本框中。

I will give you the programme of the event i am attending that has talks at 5 separate theatres running concurrently, over 2 days. The details of each talk consist of the time; title; theatre name; speaker name and designation; and for some, the description of the talk. Please rearrange and present the information given into a table form for each of the 5 theatres in chronological order, including time start, time end, title, speaker details, and the provided description if available in the information . I will give you the information for the event's first day and after your response containing all 5 theatre, the second day.
############
Big Data & AI World Keynote Theatre
Hall F, Basement 2

10:00 10:20
Augment Data Management with AI
Big Data & AI World Keynote Theatre
Sachin Tonk
Sachin Tonk, Deputy Chief Data Officer - GovTech
10:25 10:45
The Impact of Generative AI
Big Data & AI World Keynote Theatre
Introduction to Gen AI What are the potential use cases? Why does it matter to employment and business in Singapore? What are the risks? What is IMDA's role?
Dr Eric Tan
Dr Eric Tan, Assistant Director (AI Data Engineering) - IMDA
10:50 11:10
Unlocking Growth with Customer Behavioural Data
Big Data & AI World Keynote Theatre
Ben Lu
Ben Lu, Director, Solution Consulting, APJ - Amplitude
11:15 11:35
Democratizing Geospatial Data: From Data Collection to AI + Cloud Processing, and Visualization
Big Data & AI World Keynote Theatre
Key takeaways: How geospatial is collected The issues faced after the data is collected How is the cloud leveraged to store and process the large amounts of geospatial data How is the cloud leveraged ...
Kevin Lu
Kevin Lu, CTO - Skand
11:40 12:00
Innovating at 10X: Building Data & AI solutions with Amro & Domo
Big Data & AI World Keynote Theatre
Overview of Mohammed's experience at Domo and McKinsey in Building Data Solutions, including common challenges faced, and approaches to increase time to value 10x Deep dive into how AMRO built an econ ...
Mohammed Aaser
Mohammed Aaser, Chief Data Officer - Domo
12:55 13:15
... ...

结果是什么?一个美丽的计划表,提供了相关的信息,我可以用来计划接下来两天的探索。

ChatGPT中文站
ChatGPT output in table form

链接的完整对话

由于我无法在用户界面表单中调整输出令牌限制,只提供了前两个位置的表格,但通过快速提示指定缺失的输出,然后输出出来了。然后为接下来几天的节目进行重复操作,我以更熟悉的格式得到了所有信息,而且只需要几分钟时间。在这种情况下,我选择将其转移到电子表格中,这样我就可以在展览场地漫游时方便查看。

总的来说,这是一种相当快捷和简单的方式,将信息转换为更为熟悉和易于阅读的格式。同时,我测试了ChatGPT的限制,达到了其输出令牌限制但没有达到其输入令牌限制,并更好地了解了它处理未格式化但半结构化数据的能力。

更重要的是,这引出了一个观点,人工智能和深度学习模型(LLM)甚至可以在最平凡的任务中帮助我们,这些任务本来需要花费我们大量的时间和精力,现在可以轻松完成。事实上,即使采用这种简单的复制粘贴方式进入用户界面,想象一下通过利用API(正如先前提到的文章中所述)所能达到的潜力。在这个时代,不重新评估和考虑我们手头可用的不断扩展的工具来让生活更轻松、更简单、更美好是不明智的。

关于大数据和人工智能世界展览,这是一次非常棒的经历,特别是因为我可以轻松地在手机上查看在电子表格上标注的各个演讲的时间和地点。

更重要的是,地板上的时髦词是GenAI 😉

2023-10-20 17:18:01 AI中文站翻译自原文