Sanjeev Kumar

Technology Executive VP | Entrepreneur | Ph.D. AI Researcher | Author | Data & Analytics Evangelist | #DeepLearning | #MachineLearning

[Sanjeev Kumar @SanjeevK_]

The common struggle to implement successful data strategies demonstrates the need for businesses to bridge the gap between data aspirations and reality. This can be mitigated by pushing for more comprehensive approaches to data utilization and integration into #business strategy.

[Sanjeev Kumar @SanjeevK_]

Building a data-driven culture requires both top-down leadership and user-friendly tools for employees to access and utilize #data effectively. https://t.co/o5rPUQXYxR

[Sanjeev Kumar @SanjeevK_]

As data continues to be increasingly essential, there's a growing need for #DataLiteracy from a wider audience of people in roles that previously didn't require an understanding. This means knowing how to effectively collect, analyze, & interpret data to make informed decisions.

[Sanjeev Kumar @SanjeevK_]

Though #data has been an important part of business for quite a while now, the need for it has certainly grown, and utilizing it effectively can give businesses a major competitive advantage. https://t.co/4sOp2zjR9f

[Sanjeev Kumar @SanjeevK_]

➡️ Ensure that your data is accurate, relevant, and high quality. AI is all about the data, and its effectiveness relies on the quality of the input.

Keeping these 5 things in mind can set your business miles ahead of competition.

#Strategy #Success

[Sanjeev Kumar @SanjeevK_]

Adopting #AI for your #business sounds great on the surface-level, but there are many things that people fail to consider. Before you begin to fully embrace AI in your business, I urge you to be mindful of a few things...

[Sanjeev Kumar @SanjeevK_]

➡️ Prepare your workforce for AI adoption by providing training programs. Encourage collaboration between employees and AI systems for optimal results.

[Sanjeev Kumar @SanjeevK_]

Keeping ahead of the competition doesn't have to be difficult—with #AI on your side, you can gain valuable insights that will greatly improve the outlooks for not just businesses, but employees. Remember, when used right, #ArtificialIntelligence is here to augment, not replace.

[Sanjeev Kumar @SanjeevK_]

Recent studies have found that "...businesses who adopted artificial intelligence early on experienced a 100% cash flow increase," while "...if businesses don’t adopt AI, they will see a 20% cash flow decline." https://t.co/a7YLsZ6eBs

[Sanjeev Kumar @SanjeevK_]

Through it, we're seeing enhanced efficiency, reduced costs, and increased competitiveness in manufacturing, driven by its continued evolution and adoption across key areas like customer interactions, supply chain management, and workflow optimization.

[Sanjeev Kumar @SanjeevK_]

#AI is transforming #manufacturing by analyzing data and automating decisions, offering improved customer outcomes and #SupplyChain visibility. https://t.co/Mk2vU4ieLG

[Andrew Ng @AndrewYNg]

It is only rarely that, after reading a research paper, I feel like giving the authors a standing ovation. But I felt that way after finishing Direct Preference Optimization (DPO) by @rm_rafailov @archit_sharma97 @ericmitchellai @StefanoErmon @chrmanning and @chelseabfinn. This beautiful paper proposes a much simpler alternative to RLHF (reinforcement learning from human feedback) for aligning language models to human preferences.

RLHF has been a key technique for training LLMs. In brief, RLHF (i) Gets humans to specify their preferences by ranking LLM outputs, (ii) Trains a reward model (used to score LLM outputs) -- typically represented using a transformer network -- to be consistent with the human rankings, (iii) Uses reinforcement learning to tune an LLM, also represented as a transformer, to maximize rewards. This requires two transformer networks, and RLHF is also finicky to the choice of hyperparameters.

DPO simplifies the whole thing. Via clever mathematical insight, the authors show that given an LLM, there is a specific reward function for which that LLM is optimal. DPO then trains the LLM directly to make the reward function (that’s now implicitly defined by the LLM) consistent with the human rankings. So you no longer need to deal with a separately represented reward function, and you can train the LLM directly to optimize the same objective as RLHF.

Although it’s still too early to be sure, I am cautiously optimistic that DPO will have a huge impact on LLMs and beyond in the next few years.

You can read the paper here: https://t.co/m14qRYszVa I also write more about this in The Batch (linked to below).
https://t.co/8h2ag2plIa

[Sanjeev Kumar @SanjeevK_]

The past year was certainly exciting, but 2024 promises to make up for many areas where we currently lack. What are you most excited to see improvements for?

[Sanjeev Kumar @SanjeevK_]

A bit late but wanted to post, #2023 was a big year for #AI, with #ChatGPT taking off, #Bard launch, and the development of many other programs being launched to popularity, and now there's no telling what’s next in #2024.

But we can certainly speculate. Here's what I see...

[Sanjeev Kumar @SanjeevK_]

➡️ Heavier emphasis on user privacy & public trust.
➡️ Further integration into many more industries.
➡️ Increased employee training, knowledge, upskilling, data and AI literacy.
➡️ Greater adoption into startups for areas such as automation, CX, etc.

[Sanjeev Kumar @SanjeevK_]

New material found by #AI could reduce lithium use in batteries https://t.co/ZIbb1jaEJ1

[Sanjeev Kumar @SanjeevK_]

By implementing a results-driven AI strategy, businesses can unlock unprecedented opportunities for #growth, #innovation, and #success!

[Sanjeev Kumar @SanjeevK_]

Nowadays, an #AI strategy is not optional. AI can revolutionize industries, improve operations, and open new doors. But without a clear plan, businesses risk falling behind and missing out on this technology's full potential. Here are 3 simple tips to make the most of it...

[Sanjeev Kumar @SanjeevK_]

🧑‍💻 𝐈𝐧𝐯𝐞𝐬𝐭 𝐢𝐧 𝐭𝐡𝐞 𝐫𝐢𝐠𝐡𝐭 𝐭𝐚𝐥𝐞𝐧𝐭 𝐚𝐧𝐝 𝐢𝐧𝐟𝐫𝐚𝐬𝐭𝐫𝐮𝐜𝐭𝐮𝐫𝐞. Without the right technology and people, your AI initiatives will likely fail to deliver the desired outcomes and may ultimately be abandoned.