Do you really need to train your own LLM?

Today’s Question: How can you reduce the time and expense of training large language models (LLMs) for generative AI apps??

Want to get started with generative AI quickly? Of course you do. One of the first things you need to do is add large language models (LLMs) to your infrastructure to start powering generative AI applications.?

The problem is training your own LLM (despite its benefits) is time-consuming, resource-intensive, and expensive. Our SVP of technical audience relations, Christophe Coenraets , works with developers to test new AI possibilities at Salesforce. Here, he breaks down the pros and cons of three different ways to work an LLM into your day-to-day:

• Train your own, which can be daunting.
• Customize a pre-trained open source model, which is easier but still takes time and resources, including specialized engineers.
• Use an existing model from OpenAI, Anthropic, Google, and others through APIs, which saves you a lot of time and resources.

4 ways LLMs help you serve your customers better?

LLMs are masterful at recognizing patterns and connecting data on a far greater scale, and much faster than humans. From sentiment analysis to AI-driven marketing campaigns and more, here’s how LLMs help you serve your customers better.?

What our community is saying?

A problem with human languages is they're ambiguous, which is why when you prompt an LLM, you don't always get a predictable result. Whereas if you code in Python, that's a very unambiguous language to tell a computer what you want. So I think for the foreseeable future, someone that knows how to use LLMs, how to prompt them and who knows how to write a little bit of code will be able to do much more than someone that only knows how to prompt LLMs.?

— Andrew Ng , founder of DeepLearning.AI , found and CEO of LandingAI , and co-founder and chairman of Coursera , on the Ask More of AI podcast with Clara Shih .

What we’re reading?

• CIO Online looks at the role of cloud-based LLMs. LLMs, it notes, can automate threat detection and response by identifying hidden threats, and can analyze communication patterns to prevent even sophisticated phishing and pretexting attacks.?
• VentureBeat reports on a new platform, MonsterAPI, that aims to democratize access to LLMs, making them more affordable and useful and letting users fine-tune open source LLMs without writing any code.?
• Take that! MIT Technology Review reports on a tool called Nightshade, which lets artists “poison” training data by adding invisible changes to pixels in their art before uploading it to the internet. If it’s scraped into an AI training set, it can cause the model to break, and act chaotically and unpredictably.?

What’s the biggest question you have about AI? Let us know with a comment below. We read every single one and may choose yours as the subject for a future newsletter.

Get answers from experts and business leaders on trusted AI: AskMoreOfAI.com.