I was curious as to how new kid on the block ChatGPT compared against Google Translate. I gave each system 12 lists of 10 thematically linked Thai words and compared the translations against my expected translation and scored them 0, ½ or 1 according to how close they came. (This is, of course, pretty subjective.) ChatGPT scored 96/120 (80%) and Google 97.5/120 (81%). Neither is particularly impressive, and I wouldn't suggest using either as a dictionary.

Interestingly ChatGPT recognised that the blocks of words were thematically linked. When I accidentally included แว่นตา in a list of medical conditions ChatGPT responded:

แว่นตา is a word for "glasses" or "eyeglasses" in Thai. As for the other words, they refer to various medical conditions

Rather irritatingly, ChatGPT used different formats for the translations: sometimes it included the original Thai in the responses, sometimes not; sometimes the Thai was first, sometimes second; occasionally it included transliteration.

Anyway, for what it's worth, here are my results:

Notes

Some people have said they disagreed with some of my "expected translations", and said, for example, that เพลง should be "song", not "music", and เนื้อ should be "meat", not "beef". However, the "expected translations" are in the context of the theme of each of the groups of ten words. So, in the context of the groups, I stand by my expected translations.

When asked for the translations of these individual words it gave the following responses:

The word เพลง in English means "song".

เนื้อ can mean different things depending on the context. Here are some possible translations:
Meat (as in the flesh of an animal used as food): ชิ้นเนื้อ
Body (as in the physical structure of a human or animal): ร่างกาย, ลำตัว
Substance (as in the essence or core of something): สาร, สาระ, แก่นของ
Text (as in the content of a written or spoken work): เนื้อเรื่อง

ChatGPT appears to understand that the words are thematically linked when providing translations, so I wondered what ChatGPT considered the themes to be. Using the same groups of words, I asked "What do these words have in common?" and provided a list of words. The responses were very accurate:

1. they represent different academic fields or subjects
2. they relate to government and politics
3. they represent different types of buildings or establishments that provide various services to people
4. they relate to the military
5. they relate to nature, specifically to plants
6. they are all types of meat or meat products
7. they all refer to different types of rice
8. they are all related to finance or business
9. they are all accessories or clothing items that people can wear
10. they all refer to different types of medical conditions or diseases
11. they are all types of kitchen appliances or tools that are commonly used in cooking or food preparation
12. they all refer to different fields of study or areas of expertise

I would only quibble with #7 - oats and sweetcorn are not types of rice.

ChatGPT can be found at https://chat.openai.com/chat (registration required).

Google Translate is at https://translate.google.co.uk/