This past Sunday was the Five Boro Bike Tour. I was nervous about whether I would be able to ride after I finished my volunteer work for the event, due to my lack of riding this spring and gaining weight this past year. It was a little cold and windy, especially on the Queensboro and the Verrazzano Bridges.
Baruch Hashem, I was able to do it, and then I had to ride from the Staten Island Ferry to my car near the Queensboro Bridge. I enjoy the tour because I can ride on roads that I normally cannot ride on, such as the FDR, the BQE, and the Verrazzano Bridge. Although I would have liked the weather to be a few degrees warmer and less windy, I would take this instead of rain and hot weather.
I was stationed at my usual spot in Central Park. There is an area where joggers can run on the sidewalk, but there is always a battle with some joggers who do not want to move off the path where the bikes are going. It is clearly a risk for both the joggers and the bike riders to be in the same area. During the three-and-a-half hours I was working, there were 32,000 riders, many of whom came in swarms.
This year, it was not my job to deal with the joggers. In the past, when we asked the joggers to move, some would just ignore us. This year, I heard one curse another volunteer out. Another ran into the volunteer with his hands up to make it look like it was an accident. These are just two examples of how nasty and self-absorbed our society has become.
Now to my main topic. AI seems to be more prevalent day by day. Companies such as Google, Amazon, and Meta are spending billions of dollars on AI. There seems to be a feeling among many that AI is a panacea. The current situation is an indication that they are incorrect. There have been instances where attorneys, including those from large, prestigious law firms such as Sullivan & Cromwell, relied upon AI and have suffered the consequences. For example, AI created citations for cases that did not exist, which were used by the lawyer in filing a brief with the court. AI wants to please the attorney so much that if it cannot find anything, it makes it up. It is referred to as “hallucinations,” which are instances in which artificial intelligence tools fabricate case citations, misquote authorities, or generate non-existent legal sources.
I have had my own experience with AI, which involves the Queens Jewish Link. Until the paper started using AI, it was rare for my columns to be changed, since I have my own editor who is exceptionally good at what she does. Three weeks ago, I was reading my column and noticed changes to my language, some of which affected the meaning. The most blatant change was the addition of a sentence at the end of the article to connect it with the beginning of the article. AI did not understand that with many of my columns, my first topic is short. I then spent most of the column on the second topic. Under such circumstances, it would not make any sense to end the column with a reference to my first topic.
Then last week, it added a word to the sentence that clearly changed the meaning. I wrote: “In any event, those who think that switching to the Republican Party is the solution are deluding themselves.” The meaning of the sentence is clear: that switching to the Republican Party is not the answer to dealing with anti-Semitism in the United States, and those who believe so are fooling themselves. AI changed the sentence to: “In any event, those who think that switching to the Republican Party is the sole solution are deluding themselves” (emphasis added). The addition of the word “sole” changed the entire meaning of the sentence. The new meaning is that switching to the Republican Party is a solution, but you are deluding yourself if you believe that it is the only solution. There are other actions you can take. In my case, fortunately, the next few sentences indicated that I do not believe that switching to the Republican Party is a solution.
The point is that AI makes mistakes, and you must be careful in checking out AI to make sure that it is accurate, whether it is in the information provided or its changes to what you write. If not, you may learn the hard way what happens when you rely on AI and it makes a mistake.
Editor’s Note: We apologized to Warren by email and would like to publicly apologize here as well. We use AI to help with copy editing for some articles, but if the prompts aren’t clear enough, it sometimes takes greater liberties and makes unwanted changes. The changes made to this column the last two weeks were missed before publishing, and we are sorry for that oversight.
