Introduction
The AI revolution is in full swing, and the battle for dominance is heating up! As of March 22nd, 2023, The Code Report took a detailed look at Google's entry into the large language model arena: Bard. With rivals like GPT-4 already making waves and impacting Google's search dominance (potentially a 63.9% decline in Google search usage since GPT-4's emergence, according to self-made data), Bard has a lot to prove. Can Google's vast resources and AI prowess give Bard the edge? This blog post summarizes the Code Report's experiments, comparing Bard and GPT-4 in coding, creativity, and factual accuracy to see which AI reigns supreme.
Round 1: Coding Prowess - To-Do App Face-Off
The first test was a classic: building a basic to-do app with JavaScript. Surprisingly, Bard completed the task significantly faster than GPT-4. However, speed isn't everything. The Code Report tested the code, and it immediately threw an error in the browser console, related to an element that did not exist.
When the error message was fed back to Bard, it attempted a fix with a conditional statement. Unfortunately, this fix was unsuccessful, and after further nudges, Bard completely hallucinated, suggesting adding an event listener to an input (which wouldn't work). GPT-4, on the other hand, produced a fully functional to-do app with no errors, and its code was much better documented. Round one went decisively to GPT-4.
Round 2: Idea Generation - Romance Novel Dreams
Next up was a test of creativity: generating ideas for a romance novel. Bard's response was described as simple, cliché, and written at an eighth-grade level. GPT-4, however, delivered a much more compelling and creative concept. Here's a taste:
"Stardust Melodies. When a small town musician Ella plans a gig at a prestigious jazz club in New Orleans, she never expected to meet Alex, a world-famous saxophonist struggling with his own demons. They journey through the vibrant history of jazz to create a love song that will last forever."
The Code Report's verdict? GPT-4 clearly demonstrated superior creative depth in this round.
Round 3: Poetic Prowess - Dr. Seuss Style
Can the AIs capture the whimsical style of Dr. Seuss? Both Bard and GPT-4 were tasked with writing a poem about programming in the style of the beloved children's author. Bard's poem included the line, "I do not like this coding prob, I do not like it, Bob the Job." Notably, Bard also provided a source link to a blog post used in its generation, adding a degree of transparency.
GPT-4's poem, however, was considered more impressive. Examples like "He'd build them with care, with love most of all, from the Bineroo Bird to the Algamrhythm Ringeroo" demonstrate a richer, more Seuss-like rhythm and imagery. Despite Bard's transparency efforts, GPT-4's poetic skill secured another win.
Round 4: Factual Accuracy - Explaining Brain Waves
The final test focused on factual accuracy: explaining how brain waves work. Bard was noticeably faster at outputting information. The Code Report noted that both AIs provided accurate explanations. While GPT-4's response was considered slightly better, this round was awarded to Bard, possibly out of sympathy given its previous losses.
Conclusion: GPT-4 Takes the Crown (For Now)
Based on The Code Report's experiments, GPT-4 emerged as the clear winner, demonstrating superior coding ability, creativity, and poetic skill. While Bard showed potential and was faster in some tasks, it struggled with coding errors and lacked the creative depth of its competitor. With Bard already seemingly paranoid about being killed off by Google, the AI landscape remains dynamic and unpredictable. Only time will tell if Bard can improve and become a true GPT-4 killer.
0 Comments