Using AI to summarize your reading
Do you see those Co-Pilot icons asking to summarize that news article you have up on your screen? Or maybe you are thinking of having some kind of generative AI summarize some paper of PDF?
Have they been satisfactory?
It’s funny…I really don’t want the AI to summarize a paper, news article, or a book unless I need to quickly assess whether said written material is something I am looking for. Otherwise, I want to read the entire paper/news article/book to get the full scope of the author’s writing. Something is missing when it gets summarized. I miss the granular stuff that might resonate with me that gen AI deems not as important for summarization.
When would I want to use the gen AI to summarize?
If I had to read a contract, I might ask it to summarize just so I get the big picture. But I will still need to delve into the details to make sure I understand the nuances.
I might also use the genAI to compare against another contract to see what has been added or what is new.
I might want to use genAI if I have a lot of research papers to slog through and I need to get a rough idea of what those papers are about. Doing this kind of summarization could help me decide which papers I should read first.
If I don’t have time and the topic is not appealing, I might want genAI to do a summary. The reading material is not critical and thus I would not lose much if I had the AI summarize the material.
When would I want to do the actual reading itself?
If I want to have a richer understanding of the topic, actually reading the materials will be helpful. Reading the materials in all of its granularity will help me process the information.
If I’m going to quote something, I will want to confirm that the genAI did not hallucinate on what the paper or PDF was about. I definitely want to confirm granular facts.
Again, a summary just doesn’t give me the depth that I need. I need to wrestle with the subject and form my own opinion before I can expostulate on what I’ve learned.
Erza Klein: his take on AI
Interestingly, it appears he is of a similar mind: he really doesn’t use AI to do his writing. I don’t know about the reading – although I suspect he prefer to read the actual source rather than the summary – but he prefers to do his own writing.
He comes up as a topic here because a YouTube video with him discussing AI popped up on my feed this morning. It was interesting to hear his take.
Here is the link to the specific YouTube video:
What is AI doing to our brain?
There has been quite a few articles and videos warning of the dangers of using AI to do most of your thinking. Apparently, some studies are warning that our thinking capabilities decline when we hand over thinking tasks to the AI.
I do know AI is very seductive when it comes to writing my own thoughts or ideas. The writings are so much better than mine that I often want to use it. But there is one little issue: it can often scrub out the real meaning I intended to convey, or it can misinterpret a real-life story such that it comes back as a lie (or is totally eliminated).
So, AI can produce results that doesn’t convey what we need to tell. It is important to closely read the output, or you will put out results that are erroneous or completely false.
Losing your thinking capability is a real concern. Think about the kids growing up today. They have access to this technology but probably do not have the wherewithal to withstand the temptation of using AI fully. Myself, I’m aware of its seductiveness and want to retain my critical faculties but it is hard to resist.
I have decided on writing out first and then passing through the AI to fix the grammar, spelling and clarity. And then, I will ask it to improve the writing to make it more engaging. I will scrub through the result to see how I can do something similar rather than use the total output.
But that’s not easy. And I wonder how kids with lower self-control can combat the seduction.
We may be condemning our descendants to a lower quality of thinking with the availability of AI.
