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 […]
The only thing AI or machine learning algos can do better than humans or existing technology is processing volumes of data. Otherwise, AI can be equally inefficient in accuracy (probably a bit better). In case we expect to have 100% accuracy, then that’s the fault on our side. In case a system can do that, […]
* Can AI predict everything?* Are we expecting AI to predict everything?* If everything is predicted, is that AI?* If AI couldn’t predict 100% then what/who will make the decision?* Won’t be bothering if AI says a % of confirmation and leaves the remaining to the human to decide? #AI #engineeringleadership
I thought AI would learn human? If the humans needs to learn prompt engineering to use LLMs then aren’t we – the humans are learning the so called AI? 😂 #AI #llm #softwareengineering
Trust me AI is not for everyone. Finding an algo is age old thing and there are plenty at disposal BUT getting trained your model with right set of data is one heck of thing and NOT for everyone (I meant the businesses who try to implement) Even the biggies are struggling to keep it […]