Live Hindi News :- It is widely believed that creative artificial intelligence (AI) services, including Chat GBD, can answer any question. Not only this, while AI chatbots are said to be capable of surpassing human capabilities in many cases, their limitations and limitations keep coming to the fore from time to time. A recent incident is an example of this.
'Strawberry' confusion: Have you noticed that the letter 'R' appears twice in the English word 'strawberry'? The fact that Chad GBD does not know how to spell the English word 'strawberry' is currently a hot topic on the internet. When asked how many times the letter 'R' appears in the sequence of letters in 'strawberry', Chad GBD replied 'The letter 'R' appears twice. When he was told that this is wrong, he replied, 'Sorry, there are three 'R's.'
Similarly, when this question was asked to chatbots like 'GBT4', 'Clad', 'Meta AI', two of them answered it with the letter 'R'. On being asked for clarification, the 'GBT4' chatbot divided the word 'strawberry' into two parts as 'straw' and 'berry' and said that 'berry' has two 'R's'. Tech experts say that this incident is more about the need to understand how AI chatbots work than the inadequacy of AI chatbots.
What can be done? – While AI chatbots are capable of conversing with humans, they do not have the ability to understand everything. Although this training given to AI is considered machine learning, the language models (LLMs) that drive AI chatbots do not absorb and understand anything. Instead, they act and respond by converting words into ‘tokens’ and calculating the relationships between them.
This is why AI chatbots fail by giving wrong answers to even simple questions like letter order. The TechRange site has a great explanation about this. Check out that link: https://tezzbuzz.com/2024/08/27/why-ai-cant-spell-strawberry/
A year ago when Rob Manuel was asked to draw a picture of the famous band 'The Beatles' having tea, he shared his experience that Chad GBD created an amazing picture and drew a fifth picture with four members of the group. When Chad contacted GBD for clarification, he said that there was a lot of confusion about how the fifth person came in. Read more about this: Beatle-drawing
So, keep in mind that such mistakes can happen while giving inputs to AI. Now that the training for AI has improved and software development companies have started working on correcting the mistakes.
Comments are closed.