Meta's AI assistant is fun to use, but can't be trusted -bloggerheart.com


Over the past few days, you may have noticed something new inside Meta's apps, including Instagram, Messenger, and WhatsApp: an artificially intelligent chatbot.

Within those apps, you can chat with Meta AI and ask “What's the weather this week in New York?” You can type questions and requests. Or “Write a poem about two dogs who live in San Francisco.” The assistant will immediately respond, like “The Corgi was short, the butt was so wide, the Lab was tall, the tongue was slippery.” You can also instruct Meta AI to draw pictures – like an illustration of a family watching fireworks.

It's Meta's response to OpenAI's ChatGPT, the chatbot that propelled the tech industry forward in 2022, and similar bots including Google's Gemini and Microsoft's Bing AI. Meta Bot's image generator also competes with AI imaging tools like Adobe's Firefly, MidJourney, and DALL-E.

Unlike other chatbots and image generators, Meta's AI Assistant is a free tool included in apps that are used by billions of people every day, leading to a big tech company bringing this flavor of artificial intelligence to the table. The most aggressive effort to date is from the US – what is known as Generative AI. to mainstream.

“We believe Meta AI is now the most intelligent AI assistant you can use independently,” the company's chief executive Mark Zuckerberg wrote on Instagram Thursday.

The new bot invites you to β€œAsk Meta AI anything” – but my advice, after testing it for six days, is to do it with caution. When you look at it as a search engine it makes a lot of mistakes. For now, you can have some fun: its image generator can be a clever way to express yourself while chatting with friends.

A spokesperson for Meta said that because the technology was new, it might not always respond as accurately as other AI systems. There is currently no way to turn off Meta AI inside apps.

Here's what doesn't work well in Meta's AI – and what does.

Meta announced its chatbot as a replacement for web search. The company said that by typing a query to Meta AI in the search bar at the top of Messenger or Instagram, a group of friends planning a trip can see flights while chatting.

I'll say it clearly: don't do it. Meta AI fails spectacularly at looking at basic search queries like recipes, airfares, and weekend activities.

In response to my request to look up flights from New York to Colorado, the chatbot listed directions for taking public transportation from the Denver airport to downtown. And when I asked for flights from Oakland, California, to Puerto Vallarta in Mexico, the bot listed flights departing from Seattle, San Francisco, and Los Angeles.

When I asked Meta AI to look up a recipe for baking Japanese milk bread, the bot produced a generic bread recipe that omitted the most important step: tangzhong, a technique that involves cooking flour and milk into a paste.

AI also generated other basic information. When I asked it for suggestions for a romantic weekend in Oakland, its list included a fantasy business. And when I asked her to describe herself – Brian Chen the journalist – she said I worked at The New York Times, but incorrectly mentioned a tech blog I never wrote for, The Verge.

Bing AI and Gemini, which are directly linked to Microsoft and Google search engines, performed better at these types of search tasks, but clicking on a link via an old-fashioned web search is still more efficient.

AI chatbots work by looking for patterns of using words together, similar to the predictive text systems on our phones that suggest words to complete a sentence. They've all struggled with numbers.

Not surprisingly, Meta's assistant stinks at counting. When you ask him about a five-letter word starting with the letter w, he will answer with the word “wonderful”, which has four letters. When you ask him about a four-letter word starting with w, he'll tell you “wonderful,” which has three letters. Gemini and ChatGPT also fail these tests.

Like other chatbots, the more information Meta provided, the better it performed.

It excelled at editing existing paragraphs. For example, when I inserted Meta AI paragraphs that seemed verbose and asked to tighten up the paragraphs, the chatbot cut out all the unnecessary words. When I asked it to correct a sentence written in passive voice, the bot rewrote it in active voice and added more context. When I asked it to remove jargon from a paragraph written by a technical blog, it rewrote the highly technical words in simple language.

Since Meta AI is better when working with existing text, it can be helpful for study. For example, if you're taking a history class and studying World War II, you can paste a website with information about the war into the search bar and then ask the bot to quiz you. Can. The chatbot will read the information on the website and create a multiple choice test.

The most compelling aspect of Meta AI is the ability to generate images by typing β€œ/imagine” followed by a description of the desired image. For example, “Imagine a picture of a cat sleeping on the windowsill” will produce a solid image in a few seconds:

Meta's AI is much faster than other image generators like MidJourney, which can take over a minute. The results can be very strange – images of people sometimes lack limbs or look cross-eyed.

Ethics experts have raised concerns about the implications of generating fake images as they may contribute to the spread of misinformation online. But in the context of using AI when chatting with friends and family in WhatsApp and Messenger, Meta AI is a positive example of how creating fake images can be fun and safe – if we think of it as a new form of emoji. Admit it.

In a group chat with my in-laws, I mentioned that I was buying a sturdy baby stroller that could withstand the bumpy roads in my neighborhood. In a matter of seconds, my wife used Meta AI to generate an image of a stroller with giant wheels that made it look like a monster truck, with a helpful label on it that read, β€œImaginated with AI done.”

Leave a Reply

Your email address will not be published. Required fields are marked *