An OpenAI bot wrote in the BitcoinTalk forum and surprised everyone with ’smart‘ messages

The developer Manuel Araoz made an online joke to demonstrate the potential of artificial intelligence robots by having a robot write an article about itself.

According to a July 18 post on Araoz’s blog, the artificial intelligence development company OpenAI launched GPT-3, the third generation of its language prediction model capable of creating „random sentences of approximately the same length and grammatical structure as a given body of text.

AI blockchain platform creates digital assets from personal data

The blog post provides practical information on how technology can be used to impersonate familiar figures by simulating their writing styles; for example, Araoz used it to create a fake interview with Albert Einstein. He predicted that the GPT-3 could potentially replace journalists, political speechwriters and copywriters.

The phrases predicted by the robot were used in recent days on the bitcointalk.org forum, leading to a „positive“ reaction that concluded that „the system must have been intelligent.

In the blog it said:

„There are a lot of posts for GPT-3 to study and learn from. The forum also has a lot of people I don’t like. I hope they are overly enthusiastic about the possibility of having a new poster that looks smart and relevant.

Surprise!

Except that Araoz wasn’t the one writing the blog. He hasn’t posted anything on the bitcointalk.org forums for years – and he has nothing against his users. It was GPT-3 all along, he said:

„This article was completely written by GPT-3. Did you recognize it? This blog post is another attempt to show the enormous potential of GPT-3.“

According to the developer, simply providing a short biography with his information, the desired blog title and some tags was enough for the robot to create the original 750-word piece.

„I generated different results a couple of times (less than 10) until I felt that the writing style was something similar to mine, and I published it,“ said Araoz. „I think GPT-3 is one of the biggest technological advances I’ve seen so far, and I hope to play with it a lot more.

Latin America Imports ICT‘ contest opens to encourage SMEs in Blockchain, IA and Big Data technology to enter the region
AI makes predictions about blockchain

In the days before and after the blog publication, Araoz was publishing the results of his experiments with the technology on Twitter. The robot gave his opinion on blockchain, stating that it would „replace technology companies before replacing banks. Araoz was even able to get the CPT-3 to explain the work test for Iq Option reasonably well:

Tweet

It still doesn’t replace humans
The developer’s enthusiasm for this technology caused many to urgently request a test. „I would love to try something like this by training it on my own texts and see what comes out,“ said Twitter user Einar Petersen. But others reacted with fear or shock at being tricked. „I’m duly disturbed,“ said Ben Royce.

However, as advanced and entertaining as the language prediction model is, the developer does not believe it can completely replace human writers in the near future.

Alethea AI presents its project to combat the problems generated by „deepfakes
„A text-only model trained on the Internet (like the GPT-3) does not achieve intelligence at the human level,“ Araoz said. „It lacks visual understanding (e.g., nonverbal communication), complex motor skills or physical experience, and a survival instinct.