HackerNoon examined 5 AI bots on coding issues from Leetcode.com — GPT-4, GitHub Co-Pilot, Bard, Bing, and Claude+.
There’s some fascinating commentary on the strengths and weaknesses of every one — and naturally, the code that they in the end output. The ultimate outcomes?
[GPT-4’s submission] passes all checks. It beat 47% of submissions on runtime and eight% on reminiscence. GPT-4 is extremely versatile in producing code for varied programming languages and purposes. Among the caveats are that it takes for much longer to get a response. API utilization can also be much more costly and prices may ramp up shortly. Total it received the reply proper and handed the check.
[Bing’s submission] handed all of the checks. It beat 47% of submissions on runtime and 37% on reminiscence. This code appears quite a bit easier than what GPT-4 generated. It beat GPT-4 on reminiscence and it used much less code! Bing appears to have essentially the most environment friendly code to date, nonetheless, it gave a really brief clarification of the way it solved it. Nonetheless, greatest to date.
However each Bard and Claude+ failed the submission check (badly), whereas GitHub Copilot “passes all of the checks. It scored higher than 30% of submissions on runtime and 37% on reminiscence.”
