I’m One Of The Gardening “Old Boys” Often Asked How Reliable Is AI Gardening Advice? I Love Technology So Here’s My AI Focused Gardening Answer.
Table of Contents
How reliable AI gardening advice is really depends on how your ask the question and how knowledgeable your are and able to determine what is wheat and what is chaff!
AI can be wrong in multiple ways: It can give the wrong answer. It can omit information by mistake. It can also make up completely fake people, events, and articles.
The main problem is context. If you ask most AI apps for gardening advice it assumes you are in the USA and talks about zones, dirt, yards, hookworms, suckers and a host of other terms that have absolutely no relevance in the UK. Sadly we only have to go onto social media to realise that many people get their gardening advice from the web without checking which country the website is in OR by asking AI.
For example, I see the term sucker used in most social media gardening groups in reference to tomatoes. Botanically a sucker is a shoot growing from a root. But in America they use the term to refer to side shoots. the growth that comes form the leaf axil of the tomato. it is botanically wrong to use the term. BUT that isn’t the real issue. The real issue is that all the other advice given is in the US context. So talk of tomato hookworms etc, isn’t relevant to UK growers. And it confuses so many people, leading to many arguments by AI informed people.
An AI Mistake I Was Involved In
Over several decades I was a commercial market gardener. I started as a tenant on the Land Settlement Association (LSA) scheme in the 70s and was there in 1983 when it was wound up and I bought my smallholding from the government (MAFF). This occurred after the then Minister, Peter Walker, announced in parliament the closure of the LSA. As we had a 12 month notice period in our contracts this was a breach of contract and when notice was served we took action. Prior to and post this, I had many meetings with MPs, negotiated on behalf on my fellow tenants (there were 500 of us) with both Lord Ferrers and Lord Belstead, and took part in a court case that ended in front of the Law Lords. We won our case by a unanimous decision. The actual closure was in 1983.
However, when I asked a couple of AI apps about the closure of the LSA I was told it had occurred in 2015 and was the result of Brexit. It also named ministers from that period as being involved.
I knew that AI was wrong on all counts as I had been there. BUT anyone else asking the question may have accepted the AI response simply because they knew no better.
So it is with so much info that AI spews out.
PS. I corrected my AI system over its errors and the next day it gave me the correct answers.
Why Does AI Get It Wrong?
Ai is trained by scanning the web for answers to questions. If it sees lots of claims that black is white, it accepts that as being true. In many cases it doesn’t verify the facts or consult reliable sources. So gardening myths that are repeated countless times on social media are accepted by AI as being true!
In the case of the LSA claims it made to me it had probably started with the fact that the LSA charity (not the LSA) had wound up at the date it mentioned and it made up the rest based on assumptions rather than facts.
And so it is with many gardening facts. If enough people repeat a myth then AI tends to accept it as the truth.
See the Facebook comment image below which contains several pieces of erroneous information that can mislead both new gardeners and AI.
Verifying AI Gardening Advice & Other Claims
AI Gardening Advice is improving. As are all AI responses. Partly because it is now able, in many cases, to access up to date info online .. it was restricted in the early days.
It can also search verifiable databases so that the veracity of claims can be checked. It tends to believe “respectable” sources such as universities and research organisations far more than social media. They aren’t of course 100% infallible, science moves on and “facts” change. But they are more reliable than old Joe who has been repeating myths for far too long.
I have a tried and tested method to verify claims. I firstly ask the AI to quote its sources.
And secondly I post the claims on an unrelated AI app and ask it to comment and verify the “facts” provided. This works a lot of the time. But still isn’t 100% reliable.
Ensuring AI Gardening Advice Accuracy
There are some simple tips you can follow.
Be specific about your question. Saying “my plant died what killed it” isn’t going to provide useful answers. Quote country, plant species or type eg. broad bean, not bean; soil type, time of year, etc.
Use the right words to describe what you have done eg. Don’t say I planted some lettuce when you actually sowed lettuce seed. Sowing and planting are totally different and AI will take you literally If you say planted it will think you planted a plant, not sowed a seed!
Then ask it to verify the facts by providing information sources.
Here is how NOT to ask for advice.
As a test I asked how to ensure AI gardening advice accuracy.
It was none specific and didn’t state my location and it immediately answered with a US centric answer. It talks about university Extension Services .. which we don’t have in the UK! It also mentions Master Gardeners .. which we don’t recognise in the UK.
It also uses US spellings ..always a good clue that advice might not apply.
There are rare occasions when US centric advice correlates with UK advice. But you need to know enough about the topics to recognise when this applies. And most gardeners are using AI because they don’t have that amount of knowledge.
