This week, I will list twenty-one trends we observe in the industry as we shift gears from AI to generative AI.
Trend 1: ML Engineering to AI Engineering:
Just about half a decade ago, the world was building applications using machine learning models (ML engineering) to solve various business problems. Today, the whole world is developing applications with large language models (AI engineering) at a crazy speed. The number and type of such generative AI use cases are limited only by our imagination. It might take another few years to realise the true limit of generative AI applications.
Trend 2: Prompt Engineering to LLMs Generating Prompts:
The interaction between humans and large language models became necessary. Prompt engineering garnered massive interest. The quality of prompts or instructions to the LLM determined the quality of its responses. With prompting itself getting automated and augmented using LLMs, the raw skill of writing a prompt is staring at extinction. I have hardly seen a wave recede so fast.
Trend 3: CPU to CPU + GPU:
When we bought laptops or desktops earlier, we never worried about the GPU and VRAM size. Today, unless a system has a minimum of 8GB VRAM, we don’t consider it a computer. The first capability we enquire about at the store is whether it can run the deepseeks of the world locally.
Trend 4: Search to Search with LLMs:
Looking online for information on various ways Yoga removes rust from our bodies or the industry developments in generative AI over the past month? Google no more. Just LLM it. It’s not for nothing that Perplexity is my best friend and advisor these days.
Trend 5: Jobs to Jobs with Generative AI:
Let me know if you come across a job description (JD) that does not mention the need for generative AI skills. It is highly likely that the recruiter forgot to include it.
Trend 6: Traditional ML Models to LLMs:
Don’t Say a Word.
Trend 7: RNN to Transformers with Attention:
I wonder if people are still discussing recurrent neural networks (RNNs) for natural language programming tasks. They can do that only if they take a break from chanting about transformers with the attention mechanism.
Trend 8: Developer to Citizen Developer:
I have also written about citizen development on earlier occasions. Large language models have given rise to AI pair programming tools, which enable a not-so-technical user to start coding comfortably.
Trend 9: Coding to Coding with LLMs:
The LLM is the brain that does the heavy lifting behind the scenes in AI pair programming tools.
Trend 10: Manager to Tech Manager:
The middle management layer in large organisations will see a considerable shift in roles and responsibilities. People and project managers must technically equip themselves and be technical managers for their teams. And guess who is there to help? The large language models.
Trend 11: Dumb to Intelligent (or the reverse?):
I would love to say a few words, but I can’t find any. I feel a little dumb. Is it because of all the chatting I do with LLMs?
Trend 12: Training Data to Lack of Training Data:
I would never have imagined that the Internet, replete with content, would not supply the data needed to train large language models. With newer restrictions in place from content owners, we are running out of original content of superior quality for training future LLMs.
I will only mention without explaining the rest of the generative AI trends that I see in the industry.
Trend 13: T to V to Comb shaped Skillset. Trend 14: Salary to Bloated Salary. Trend 15: India in AI Engineering to India in Foundation Models. Trend 16: Regulatory Needs to Urgent Regulatory Needs. Trend 17: Electricity Consumption to Heavy Electricity Consumption. Trend 18: Core to AI Courses in Academics. Trend 19: Lack of AI Talent to Severe Lack of Gen AI Talent. Trend 20: Human-Generated Content Online to LLM-Generated Content Online. Trend 21: Python as a Programming Language to English as a Programming Language.
Facebook
Twitter
Linkedin
Email
Disclaimer
Views expressed above are the author’s own.
