Anyone who has been following Meta knows that it has already publicly debuted (and withdrawn) a chatbot. That chatbot’s name was Galactica, and its ostensible purpose was to provide researchers with assistance.
However, trouble was afoot when it turned out that Galactica was not always factually accurate in its responses. This is a common problem for many ambitious, “ask me anything” chatbots that are meant to have high-level discussions with human users.
But now, Meta has announced, via a Facebook post by founder and CEO Mark Zuckerberg, that the company has a new chatbot to offer researchers, one that is called LLaMA. No caps lock problem here, that is simply how you write it. It stands for Large Language Model Meta AI.
Meta has developed different versions of this chatbot. The most sophisticated one has 65 billion parameters, which refers to calculations in an AI algorithm that provide various “weights” to data.
So, for example, if LLaMA is indeed a quality platform, then the phrase “Force equals mass times acceleration” will have a greater weight than “Force equals mass times velocity”. The former is true, and so deserves more weight, whereas the latter is false, and so deserves less weight.
The Problem of Misinformed Chatbots
Though some people may see these chatbots’ errors as limits to chatbot technology, these errors can be put into perspective when you realize that this new wave of chatbots is still, well, a new wave.
Think of previous-generation chatbots, like Amazon’s Alexa or Apple’s Siri. When you ask either of these platforms a deep or challenging question, they tend to say “Here’s what I’ve found” and offer you links to webpages rather than a newly generated response that succinctly answers the questions.
AI developers are looking to get over the hurdle of designing chatbots that do not stop at simply directing you to a source of information. The goal now is to have chatbots summarize and explain the desired information. The content must be clear and accurate.
The challenge is compounded because AI developers are also designing chatbots that are meant to be general-purpose information fonts.
These ambitious and concurrent goals—i.e., creating chatbots that both summarize accurate information and are conversant in a wide range of topics—mean that there will be a lot of troubleshooting over the next few years with such technology.
However, once the big issues are sorted out, the payoff will be enormous. For instance, these chatbots are already being tested in search engines, which will change the future of search as we know it.
Will LLaMA Be Available for Businesses?
In these early stages, Meta is looking to employ LLaMA for “open science” initiatives. That means that LLaMA will be available to scientific researchers that are looking for assistance.
LLaMA’s role in the lab will be offering accurate information that a researcher may need throughout an experiment or analysis.
As of right now, it is uncertain whether Meta plans to extend LLaMA to businesses in general. But it would be shocking if Meta did not down the line.
Meta has specified that the platform is available for “industry”, but businesses must go through a highly selective application process to use the platform.
However, even if Meta offers complete open access to LLaMA, you may still want to go with a more reliable, industry-tested chatbot.
IBM Watson Is Available for Your Business
For now, businesses are much better going with IBM Watson, which is employed by a wide variety of industries.
Watson is a conversational computing platform that does not suffer from the pitfalls of the know-it-all “new wave” chatbots, because Watson is trained for more specialized conversations.
This way, Watson is less prone to answer incorrectly, and more likely to have an efficient, clear, and streamlined conversation that relates directly to your business’ services and products.
GO AI to Grow Your Business
Predicting the prices of raw materials with 99% accuracy. Finding out your customer churn rate well in advance. Analyzing your marketing materials to find out where viewers’ eyes will go, and in what order.
Guardian Owl Digital knows that AI is not the future—it is right now. And right now, businesses everywhere are making use of their data with AI platforms.
GO AI is our artificial intelligence initiative that offers you cutting-edge AI tools through our partnership with Findability Sciences.
We offer everything from automated chatbots that can converse with your customer base, to advertising programs that use computer vision to optimize your creative marketing strategies.
Let’s GO.
Previous GO AI Articles
For continuing your AI education and keeping up with the latest in the world of AI, check out our AI blog:
New Year, New AI: Here Are the Biggest Trends in AI Coming in 2023
How AI Could Have Helped Southwest Avoid Its Holiday Disaster
IBM Watson vs. Microsoft’s ChatGPT: The AI Chat Matchup of the Century
AI on the Stand: Explaining the Lawsuit Against the Microsoft Automated Coder
AI and You: What Determines Your AI Recommendations in 2023?
How AI Could Have Foreseen the Crypto Crash—(It Already Analyzes Exchange Markets)
Google’s Response to ChatGPT: What the Tech Giant Is Doing to Improve Its Own AI Efforts
Recent Comments