Why is this important: OpenAI released GPT-4 this week, an update to its popular language model and technology that aims to improve accuracy and is designed to act as an underlying engine for chatbots, search engines, online tutors, etc. GPT-4 is now available for paid subscribers and there is a waiting list to use the model through the API. Plus, the race for AI is on, with “AI startups” raising money like there’s no tomorrow and big tech companies like Google scrambling to let it be known that they’re not. not so far behind.
GPT-4 has been in development for most of the past year, following the quiet release of GPT-3 in mid-2020 and then ChatGPT taking the world by storm late last year. The recent rise in popularity of ChatGPT and all AI been meteoric, to say the least. The AI platform in January hit 100 million users after just two months of public availability. OpenAI used a refined model for ChatGPT which they called GPT-3.5.
NOW GPT-4 improves on that, and it’s now a large multimodal language model (MLLM), meaning it can respond to both text and images. GPT-4 learns by analyzing huge amounts of data on the Internet and has many applications for businesses, including automating the work of paralegals and content moderators on the Internet. However, like its predecessor, it has human limitations that can still present a challenge. The language model may pass some standardized tests, but fail on simpler queries.
ðÂ¤¯ðÂ¤¯Well that’s another thing.
GPT-4 passes virtually all exams. And don’t just pass…
Bar exam: 90%
GRE Quantitative: 80%, Verbal: 99%
Every AP, the SAT… pic.twitter.com/zQW3k6uM6Z
— Ethan Mollick (@emollick) March 14, 2023
Despite this, OpenAI’s GPT-4 has already attracted interest from companies such as Morgan StanleySalesforce, Duolingo, and online education company Khan Academy. Open AI and Microsoft already have a close relationship, with the software giant having invested more than $13 billion in the organization over the past few years, securing them exclusive licensing rights to OpenAI technology. Concrete example, if you used Bing Chat for the past few weeks you have literally beta tested GPT-4 before the official release.
OpenAI acknowledges that GPT-4 remains flawed and limited, but the update represents a significant step up in its capabilities. The model is more creative than before and is able to learn usage patterns and style text to match the desired result. GPT-4 can understand images, use them as input, describe them as text or also contextualize them based on them.
Imagine if you could synthesize a document you wrote into a presentation in seconds.
One of the many features we have integrated @magicaltome powered by GPT-4. See you soon. pic.twitter.com/2m3RgKcM06
—Henri Liriani (@hliriani) March 14, 2023
Good news, we have increased our turn limits to 15/150. Also confirming that the next-gen model that Bing is using in Prometheus is indeed OpenAI’s GPT-4 that they just announced today. Congratulations to @OpenAI crew. https://t.co/WTVlVCVOyw pic.twitter.com/VA4Z1SDiEG
— Jordi Ribas (@JordiRib1) March 14, 2023
Although OpenAI isn’t interested in explaining how GPT-4 really differs or how it has improved internally to work better, it is measurable. more capable as it can have longer conversations, taking around eight times more text than ChatGPT, from around 3,000 words in the previous version to handling up to 25,000 words in its latest revision.
GPT-4 is now available for users who pay for ChatGPT Plus (or via Bing Chat), while API access is granted to developers on the OpenAI waiting list.
The reactions were immediate. Check out some of the most interesting apps and related discussions around this announcement for further reference:
DoNotPay is working on using GPT-4 to generate “one-click lawsuits” to sue calling bots for $1,500. Imagine receiving a call, clicking a button, the call is transcribed, and a 1,000 word lawsuit is generated. GPT-3.5 wasn’t good enough, but GPT-4 handles the job extremely well: pic.twitter.com/gplf79kaqG
—Joshua Browder (@jbrowder1) March 14, 2023
I don’t care if it’s not AGI, GPT-4 is an incredible and transformative technology.
I recreated the Pong game in less than 60 seconds.
It was my first try.
Things will never be the same again. #gpt4 pic.twitter.com/8YMUK0UQmd
— Pietro Schirano (@skirano) March 14, 2023
We are excited to introduce Virtual Volunteer™, a digital visual assistant powered by @OpenAIof the GPT-4 language model. Virtual Volunteer will answer any question about an image and provide instant real-time visual assistance within the app. #Accessibility #Inclusion #CSUN pic.twitter.com/IxDCVfriGX
– Be My Eyes (@BeMyEyes) March 14, 2023
GPT-4 makes drug discovery.
Give him a currently available drug and he can:
– Find compounds with similar properties
– Edit them to make sure they are not patented
– Buy them from a supplier (even by sending an e-mail with a purchase order) pic.twitter.com/sWB8HApfgP
— Dan Shipper ï¿½”ï¿½ (@danshipper) March 14, 2023
Hey gpt4, make me an iPhone app that recommends 5 new movies every day + trailers + where to watch.
My ambitions grew as we went ï¿½’ï¿½ pic.twitter.com/oPUzT5Bjzi
— Morten Just (@mortenjust) March 15, 2023
Can GPT-4 encode an entire game for you? Yes, yes it can.
— Ammar Reshi (@ammar) March 14, 2023