Hold onto your hats, data lovers! Because here’s the thing: the better the data, the stronger the AI, right? We’re talking top-secret personal info, business goldmines – anything that’s valuable and true gets the most guard dogs. See, the more valuable and accurate the data, the tighter we clamp down to keep it safe, whether […]
If you closely observe how LLMs are evolving, you’ll gain immense respect for yourself. Yes, I’m talking about YOU – the human. The latest generation models can handle up to 128k tokens in a single prompt. This number might seem overwhelming, but the ability to process more tokens increases the machine’s probability of understanding context. […]
This might sound odd, but actually, that’s what we needed. After all, humankind is biased, isn’t it? We all take stands and be on a side. Machines are expected to be like that too. Now, the challenge is that the machine may be expected to represent both sides—that’s the curse for the machine. Unfortunately, humankind […]
Do you often hear that the lack of data, or the absence of quality data, is an excuse not to implement AI? Quite a lame excuse, isn’t it? In today’s business landscape, implementing AI solutions isn’t just an option; it’s a necessity. If you’ve been lagging in data recording, it’s not too late to embark […]
After the emergence of general AI (Gen AI), everyone is scrambling to fit it to the right business problems. However, most are still stuck with glorified chatbots, while some claim expertise in merely chaining prompts. In this rapid race of AI advancement, we shouldn’t be searching for AI’s use cases to fit in. Instead, we […]
LLMs’ training data cutoff is a concerning disclaimer that sometimes shakes our confidence. Some detractors use this as a reason to criticize LLMs (At times, even I have said so ;)). However, we shouldn’t worry too much about it. What standard are we using to evaluate these models? The universal benchmark seems to be Google. […]
AI can’t be a representative of everyone’s Intelligence. There is a huge debate on Google’s AI model Gemini’s capability. I am sure at the scale of Google, they have better hands to fight on their own battles. Let’s talk about the expectation on the AI models. Aren’t we expecting too too much on the AI […]
Yes, you read that correctly. The essence of what AI does involves discrimination. In today’s IT industry, anyone not discussing AI is treated as if they do not exist within the sector. Consequently, everyone talks about it, but not everyone fully comprehends it. Although AI has existed in various forms for decades, Large Language Models […]
Yes, AI can turn garbage data into gold. All you need is…well, garbage! Today, AI algorithms are becoming easier to interface with. However, the challenge lies not in the algorithms themselves, but in the inadequate data that makes them underperform. Interestingly, most businesses lack the data necessary to do anything meaningful with AI. Some do […]
Gen AI hallucination is in a way represents human intelligence. Why hallucination happens? Insufficient or Biased Data training Overfitting – data variance deprivation Complex Input Patterns – In simple words, a complex, unpredictable input/question to the model. It’s like how some of us would have attended the school/college exams without enough preparation. Model limitation and […]