As the AI race unfolds, OpenAI keeps the lead and unveils GPT-4
Meanwhile, big tech companies are scrambling to make it known that they are not so far behindBy Julio Franco 15 comments
Why it matters: OpenAI launched GPT-4 this week, an update to its popular language model and technology that aims to improve precision and is designed to act as an underlying engine for chatbots, search engines, online tutors, and more. GPT-4 is now available to paid subscribers and there's a waitlist to use the model via API. Furthermore, the AI race is on, with "AI startups" raising funds like there is no tomorrow and big tech companies like Google scrambling to make it known that they are not so far behind.
GPT-4 has been in development for most of the past year, after GPT-3 was quietly released in mid 2020 and then ChatGPT took the world by storm late last year. The recent rise in popularity of ChatGPT and everything AI has been meteoric, to say the least. The AI platform in January reached 100 million users after just two months of public availability. OpenAI used a fine-tuned model for ChatGPT which they called GPT-3.5.
Now GPT-4 improves upon that, and it's now a multimodal large language model (MLLM), which means it can respond to both text and images. GPT-4 learns by analyzing huge amounts of data from the internet and has many applications for businesses, including automating the work of paralegals, and moderators of content on the internet. However, as with its predecessor, it has some human-like limitations that may still present a challenge. The language model can ace some standardized tests, but still go wrong on simpler queries.
ðÂ¤¯ðÂ¤¯Well this is something else.— Ethan Mollick (@emollick) March 14, 2023
GPT-4 passes basically every exam. And doesn't just pass...
The Bar Exam: 90%
GRE Quantitative: 80%, Verbal: 99%
Every AP, the SAT... pic.twitter.com/zQW3k6uM6Z
Despite this, OpenAI's GPT-4 has already attracted interest from companies such as Morgan Stanley, Salesforce, Duolingo, and online education firm Khan Academy. OpenAI and Microsoft already have a close relationship, with the software giant having invested over $13 billion in the organization in the last few years, getting them exclusive rights to license OpenAI's technology. Case in point, if you have used Bing Chat in the past few weeks, you have literally been beta testing GPT-4 ahead of the official release.
OpenAI acknowledges that GPT-4 remains flawed and limited, however the update represents a significant step forward in its capabilities. The model is more creative than before and is capable of learning patterns of use and text style to match the desired output. GPT-4 can understand images, use them as inputs, describe them as text, or also contextualize based on them.
Imagine if you could synthesize a doc you wrote into a presentation in seconds.— Henri Liriani (@hliriani) March 14, 2023
Just one of the many capabilities we've built in @magicaltome powered by GPT-4. Coming (very) soon. pic.twitter.com/2m3RgKcM06
Good news, we've increased our turn limits to 15/150. Also confirming that the next-gen model Bing uses in Prometheus is indeed OpenAI's GPT-4 which they just announced today. Congrats to the @OpenAI team. https://t.co/WTVlVCVOyw pic.twitter.com/VA4Z1SDiEG— Jordi Ribas (@JordiRib1) March 14, 2023
Although OpenAI is not interested in explaining how GPT-4 really differs or how it's improved internally to work better, it is measurably more capable as it can have longer conversations, taking in about eight times more text than ChatGPT, from around 3,000 words in the previous version to processing up to 25,000 words in its latest revision.
GPT-4 is now available to users who pay for ChatGPT Plus (or via Bing Chat), while API access is being granted to developers on OpenAI's waitlist.
Also read: ChatGPT was made possible thanks to tens of thousands of Nvidia GPUs, which Microsoft is now upgrading
Reactions have been immediate. Check out some of the more interesting applications and related discussions around this announcement for further reference:
DoNotPay is working on using GPT-4 to generate "one click lawsuits" to sue robocallers for $1,500. Imagine receiving a call, clicking a button, call is transcribed and 1,000 word lawsuit is generated. GPT-3.5 was not good enough, but GPT-4 handles the job extremely well: pic.twitter.com/gplf79kaqG— Joshua Browder (@jbrowder1) March 14, 2023
I don't care that it's not AGI, GPT-4 is an incredible and transformative technology.— Pietro Schirano (@skirano) March 14, 2023
I recreated the game of Pong in under 60 seconds.
It was my first try.
Things will never be the same. #gpt4 pic.twitter.com/8YMUK0UQmd
We are thrilled to present Virtual Volunteer™, a digital visual assistant powered by @OpenAI's GPT-4 language model. Virtual Volunteer will answer any question about an image and provide instantaneous visual assistance in real-time within the app. #Accessibility #Inclusion #CSUN pic.twitter.com/IxDCVfriGX— Be My Eyes (@BeMyEyes) March 14, 2023
GPT-4 does drug discovery.— Dan Shipper ï¿½"ï¿½ (@danshipper) March 14, 2023
Give it a currently available drug and it can:
- Find compounds with similar properties
- Modify them to make sure they're not patented
- Purchase them from a supplier (even including sending an email with a purchase order) pic.twitter.com/sWB8HApfgP
hey gpt4, make me an iPhone app that recommends 5 new movies every day + trailers + where to watch.— Morten Just (@mortenjust) March 15, 2023
My ambitions grew as we went along ï¿½'ï¿½ pic.twitter.com/oPUzT5Bjzi
Can GPT-4 code an entire game for you? Yes, yes it can.— Ammaar Reshi (@ammaar) March 14, 2023