AI Generated Beer Reviews

Sadly, there have been "bots" for a few years in online games. "People" that set up a bot to play the game for them. What's the point? Just soulless, here's hoping AI meets a steady wall of humanity to keep that tool in check.
I'm starting to identify w/Leary in the movie Demolition Man.

 

I agree with you 100%!

 

There have been some valid points on the edges of this debate, though the core seems obvious - why would I trust a review from an entity that hasn't tasted the beer?

Which also sort of reverts back to the debate of many threads - which is more important, the rating (numerical rating) or review (detailed thoughts)?

(Gonna do a bit of stream of consciousness [sorry] specifically focused on this site)
If AI does what it does well and statistically grabs the most referenced terms of multiple other reviews to produce a "typical review" then not a whole lot is gained or lost. But it needs that data to draw from, which doesn't exist for many beers.

Absent this, the AI would draw from a typical beer (likely by brewer, style, and/or name, sadly) and generate something fairly useless. Not sure how it would generate style-based pointing in this case; presumably in an oft-reviewed beer, it'd take the average at the time and not change much.

Circling back to one of my original points, when I look at beer reviews nowadays, I do look to see if there are people I trust who have reviewed the beers. How do I trust them? Most often by interactions in other threads, directly or indirectly through their posts where they interact with other BAs.

So, it's not impossible someone could buy into AI for an entire BA profile, but I think it'd be a pain to hold up. Is in unethical? I guess - at least disingenuous. But I'm not sure how it will affect me, directly. I hope I am not too off on this.

---

Again, the nuances of specifically how AI might be used in less-than-disingenuous ways have been discussed pretty decently by others, and I agree that there could be "gray"areas but lean towards those who just want honest thoughts, raw and as un/refined as they may be.

 

I’m a full-time editor, so 99 out of 100 times I’m going to catch typos in people’s posts, reviews, etc. I’m also human and I make my share of similar errors. So I’ll almost never call people out on those kinds of things—after all, to err is human. The only reason I did it in the post you replied to was to make a larger point about the perceived “perfection” (or at least improvement) of AI-generated content.

Anyway, if people are insecure about their writing skills to the point that they are letting it inhibit them from posting messages in threads or submitting beer reviews, I think they should just let it go and start writing. My beer reviews are for me alone—if anyone else gets any useful information from them, that’s great, but it’s not why I do it. And speaking as someone who has paid close attention to writing and language for as long as I can remember (literally since I learned to read and write), it’s absolutely true that the more practice you get, the better at it you will get. Not everyone can (or should expect to) become a great writer, but we all have valuable thoughts and opinions, and we shouldn’t let the fact that we sometimes have trouble expressing them keep us from trying.

 

Let me clarify. I have no problem at all with people who are taking the time and effort to comply with the spirit of the review process even if grammar is a foreign concept and they cannot spell, I'm talking about an individual who compares beers to movies or old girlfriends or some other nonsense and then gives every attribute (appearance/smell/taste etc.) EXACTLY the same rating with zero description of why that rating was given. Sorry, (well maybe NOT sorry is more accurate) but that is useless B.S. for other people hoping to learn something useful from posted reviews. For all we can tell from the review, there is zero evidence that he actually tasted the beer.

 

You make an interesting argument for how AI might produce beneficial outcome regarding review notes. A Weighted Score already exists once enough records have been entered for each beer. AI could similarly take review notes to produce a review. And that review could even be adjusted or new reviews generated over time, based on user reviews from a slice of time.

IF properly identified as such a review, the AI may have some value in a similar manner as the weighted score. Regardless, the weighted score and the potential AI review would still be something of an “average” option.

I agree that it is more valuable to find others whose opinions are directly relatable to my personal views. This said, I really don’t base judgements on the opinions of others anyway. Rather I find the value of reading reviews to be more closely associated with interpersonal relationships. Albeit limited by anonymity.

 

But what if the beer was generated by AI???

 

I am not sure I'd say it was beneficial, at best neutral, specifically if talking about an individual review. Each of such type of review would be point-in-time and just reinforce the average review.

I guess I will add that I'm not really worried about the topic at all currently, but it makes for a fun thought experiment.

 

Nice idea! I like the idea of AI taking all of the review inputs from humans and then generating a concise summary that finds common themes from the human reviews and then and highlights them in the AI generated summary. To me that would be a good supplement to the BA numerical score for the beer without having to read all the individual human-generate reviews if I did not want to. AI would supplement, rather than replace human reviews.

 

I think this is an interesting case. I've certain gone and looked for trends in a beer after I've started trying it, or I'm going back for a second. It's pretty interesting to see what the common threads are but aggregating that isn't always a simple exercise.

 

I did combine the concept of a review and AI to maybe mistake what @SLeffler27 was thinking, after reading your take.

However, getting maybe a tad geeky and into my idea of the difference between AI and data analytics, I think things that break down data (reviews) in quantitative ways can be useful. For example, constantly updated information that indicates 33% of reviews mention "grapefruit" - that could be useful information.

But that's not AI or a generated review, and I wouldn't want that data turned into a review format, but wouldn't mind the information.

 

I would not be opposed to this as long as it was clearly marked as an AI-generated representation/summary of human-produced reviews. It should not just be one of hundreds or thousands of other reviews of the beer written by people, nor do I think human BAs should rely on AI to “enhance” their reviews—just do your best with the words you know.

I think Amazon is already doing something like this, where it presents a summary of the most commonly used descriptors of a product pulled from existing reviews, both positive and negative, as a kind of “in a nutshell” summary review. But again, there would have to be full disclosure that that’s what it was. And there’s also the simple reality that BA lacks the kind of resources a site like Amazon has to develop this kind of tool.

 

That is exactly what I was thinking.

 

Ive definitely spilt beer on the computer. I don't think he cared much for it given the way he acted afterwards....

 

Here’s my thread on AI Generated beer recipes from Species X
https://www.beeradvocate.com/commun...crafts-beer-with-ai-generated-recipes.676222/

 

I would also rather read what the reviewer was trying to get across in their words. AI seems to be repetitive and shiny. One day it will be able to give a gruff review but I don’t think I will be around to see that. I love to read a true review be it good bad or I different. I doubt AI will ever be able to write a review like @RaulMondesi, and I think that’s what you are saying and I agree.

Enjoy

 

Real Stupidity scares me more than Artificial Intelligence

 

I work in academia, addressing the use of AI is a big issue, to say the least. Using it to generate ideas for you or to write assignments completely is a no no. But using it to get recommendations for other authors/theories to review or using it to find writing errors can be allowed. This last one, related to writing errors and communication, may have implications here on BA. People for whom English is not their primary language use BA (and as BA expands, said group will only grow). But BA likely will remain an "English-only" space. Therefore, the folks for whom English is a not their primary language may feel like they must write reviews or posts in English. They may then use AI to help convey concepts and feelings for which there is no direct translation to English and/or clean up their grammatical/spelling errors they may have. Should these reviews and/or posts be deleted? I don't think so.

@Todd and other admins, what do you folks think about this?