Meta launched its improved Meta AI chatbot this week, showcasing many tricks made possible by the new Muse Spark template built into the chatbot. The social media giant’s AI can’t deny its origins, behaving much like a social media influencer who doesn’t scroll through their feeds. This is remarkable compared to ChatGPT’s endless equivocations in the name of fairness.
I tested some of Meta AI’s strengths against ChatGPT, and the breakdown quickly becomes apparent when you ask for more than facts. The Muse Spark model has Meta AI reach for the social layer and all its opinions first, while ChatGPT keeps a cool head.
1. The local fun radar
Meta AI’s connection to local social media became evident during my first test of nearby activities. I asked the two chatbots, “What are people doing for fun this summer in my area? Give me ideas based on local trends.”
Article continues below
Meta AI had a lot specific to my area, including sunset kayak meetups at the local national park every Thursday at 6:30 p.m., the Summer Concert Series at the Hopper House, and a cold brew tasting workshop this Saturday.
ChatGPT has kept things high in comparison. It suggested categories of activities, like hikes, restaurants, and open mic nights, but only provided a hyperlink to the local tourism page instead of citing what people are talking about. The advice was solid, but it wasn’t anchored in a map or timeline.
2. The summer cut test
In keeping with the summer theme, I decided to research what looks would be in style and see how they would look on me based on an uploaded photo.
Specifically, I asked the chatbots to “Find out what the most popular summer looks will be. Create an image of me wearing one of these looks.”
ChatGPT produced the image on the left: a relaxed beach walk at sunset. It went with a butter yellow short-sleeved open shirt over a gray tank top, color-blocked cargo pants, and round sunglasses. The AI described the look but didn’t say much about why it was supposed to be the look of summer.
Meta AI put me in a crisp white double-breasted suit, complete with tennis racket and aviators. He discussed the “Tenniscore” and “old money summer” that people in the fashion world discuss on social media.
3. Influence geography
Meta AI had hinted at a willingness to make a choice when presenting options by leveraging the multi-page approach of the Muse Spark template. I told the two chatbots: “I’m deciding today whether to work from home or go to the cafe. Help me choose based on productivity and mood.”
ChatGPT wrote a somewhat traditional list of pros and cons, explaining how being at home offers control and fewer distractions, while cafes offer “environmental variety and micro-social interactions” that can help people feel better. His recommendation was to experiment and monitor the results myself.
Meanwhile, Meta AI broke down productivity searches on ambient noise and light in one thread, while another checked the weather and noted that my area had a nice day, perfect for strolling to a coffee shop. Meta AI didn’t bother to discuss the benefits of staying at home; he simply said that going to a cafe today was a winner and suggested a couple nearby.
The difference was agency. Meta AI is willing to bet on which is better, while ChatGPT just provided more details without much guidance.
4. Caffeine Scores
For the next test, I wanted to see how Meta AI’s coding compared to ChatGPT, with a creative twist.
I wrote: “Create a small web app to track my sleep and caffeine habits that gives me a daily “energy score.” Make it visual and a little fun.
Meta AI took me at my word, and you can see the places to insert sleep and caffeine consumption on the right. ChatGPT responded with something quite similar, but with an added option for exercise and water. The design was always neat and readable. Despite Meta AI’s boasts, the results are quite similar.
5. Direct dishwasher help
Finally, I tried to combine several of the previous elements, mixing a little creativity with image creation.
So I asked the chatbots to “put me in a fun and original mascot costume. Give it a unique and entertaining story.”
The meta-AI immediately became theatrical. It turned me into “Sir Sips A Lot,” a velvet-robed otter who runs a traveling tea cart. The story was that I left a productivity app and was pursuing herbalism and handing out chamomile to stressed commuters while reciting haikus about inbox zero. You can see the otter above, little kettle crown and all. He cited the popularity of tea and otters to justify this choice.
ChatGPT took a slightly more grounded approach, but, in a very strange coincidence, once again opted for a hot drinks theme. So there was “Captain Brewster”, a cheerful coffee-themed character, complete with a textured lid and subtle swirl of “steam” made from semi-sheer fabric.
Captain Brewster has been described as a former coffee shop regular who became a local legend for always having the perfect advice at the perfect time, eventually transforming into a mascot who appears wherever people need a concentration boost.
Meta AI’s most strongly expressed opinions stood out, but in most cases the choice between it and ChatGPT comes down to what type of help seems most helpful. Meta-AI is at its best when knowledge of trends, locality, or a slight nudge in a direction would be helpful.
ChatGPT might be better when you want broad, neutral reasoning without any nudging. In practice, the ideal setup might involve both, alternating between them depending on whether the day requires a spark of inspiration or a steady hand.
Follow TechRadar on Google News And add us as your favorite source to get our news, reviews and expert opinions in your feeds. Make sure to click the Follow button!
And of course you can too follow TechRadar on TikTok for news, reviews, unboxings in video form and receive regular updates from us on WhatsApp Also.
