Science fiction has always tried to show off a romantic view of our future. Still, often, TV shows and movies like Star Trek, Star Wars, The Terminator, and 2001: A Space Odyssey have hinted at a darker side of technological progress. Often, the darkness has stemmed from technologies like artificial intelligence. And over the last two decades, we have seen AI mature.
In the late 1990s, we saw the IBM supercomputer, Deep Blue, defeat one of the greatest chess players of all time, Gary Kasparov. We witnessed the rise of Google and its bag of AI tricks, which propel everything from ‘Search’ to ‘self-driving’ cars that are now part of its subsidiary Waymo. In the last decade, Google’s parent company, Alphabet’s other subsidiary, DeepMind, has created AlphaGo, an AI that can beat the best ‘Go’ players at the ancient game. As AI has matured, there have naturally been concerns about how this technology is regulated and deployed.
The latest rage comes in the form of ChatGPT and DALL.E 2 by OpenAI. Yes, the very same OpenAI that is backed by none other than Elon Musk. It has managed to acquire one million users in just five days at the beta stage – faster than Twitter, Google, Zoom, and even Instagram.
In the last few weeks, ChatGPT has taken the internet by storm as it’s a conversational system that can throw up information in the form of a valid query based on the question it has been asked. Its conversational knowledge base is immense, even though a wee bit dated till around 2019. ChatGPT can be creative and organised with its answers because it understands natural language like no other AI agent we have seen.
Also read: Gamers, editors, developers, designers—An Intel comeback is taking the bite out of Apple
Why is ChatGPT interesting?
On a certain level, it can replace a low-level content writer with utter ease because of its speed, grammatical accuracy, and the information that it provides at the end of the day. It can also remember the questions you have asked, and if you point out that it has made a mistake, it will remember and correct the error.
It can perform tasks that many people can’t do. For example, it can tell you how to write an app for the iPhone. It can debug code. It can show you examples of code bases for an app that will have certain specific purposes, for example, a ticketing gateway for concerts. You can ask it how to edit videos on Final Cut Pro and it will give you detailed instructions.
It will give nuanced answers to a question like “why Led Zeppelin’s Stairway to Heaven is considered the greatest song of all time”. Similarly, it will have different answers for similar questions — for example — Who is the greatest football player of the 21st century? will have a vague answer to “Who is statistically the greatest football player of the 21st century?” Will have more factual information and there will be no subjectivity in play.
It is like the ultimate research assistant in a way, where tasks that would’ve taken hours are now done within seconds and that too by something that’s not human.
In a way, this is the promise Apple made in 2011 when it unveiled Siri to the world. You ask Siri a question, and it responds with an answer. This is what Amazon and Google have also been promising with Alexa and Google Assistant, and in a way, they have access to more data, but the dream hasn’t been realised. While ChatGPT is text-based and doesn’t have a voice assistant, it has an impressive understanding and can answer questions in detail.
Many writers are getting worried that ChatGPT could have the potential to replace them. In an era where writing standards are extremely low, these concerns are valid. But in its current form, ChatGPT can at best be used as an augmentation and an added tool in the arsenal of a writer for research purposes. It cannot replace good old-fashioned research, and its text certainly lacks ‘soul’.
But if this technology keeps progressing at the same rate through the decade, chances are many rudimentary and repetitive tasks will be taken over by something like ChatGPT.
Also read: Apple revolutionised smartwatch. Then came the budget wave and more—AmazFit, Fitbit to Samsung
DALL.E 2 is arguably even more powerful
The DaLL.E 2 image generator can create realistic images or art, or graphical designs with the input of text. Again, OpenAI’s natural language processing prowess is at the forefront. This is a harder task, even though advanced graphic design tools like Canva can be used by most people.
For example, you can ask to create a photo of a ‘canine astronaut with a balloon on Mars’, a photo of ‘Lionel Messi with the football world cup trophy in outer space’ or ‘a cyberpunk photo of the Burj Khalifa’. And boom! The AI, which is running off the Microsoft Azure cloud computing platform, does its magic.
Now, to create something of these queries will take 30-40 minutes for even an experienced designer, with ideation kept apart. But DALL.E 2 can do this in less than 30 seconds. This is the future of digital art. Then it can create queries of things that don’t exist. For example, if you ask for a photo of a ‘117-year-old Labrador Retriever’ – it will produce something that is realistic.
Of course, the overall quality of the imagery is still not up to par for many professional use cases. Pixel peeping will reveal the gaps in the quality of content that DALL.E 2 delivers.
The higher running order is that these modern technologies are becoming extremely adept, and they will soon be able to take on low-level, rudimentary jobs that don’t involve a lot of skill. This will not happen in 2022, but it will likely become our reality by 2030.
In fact, Google has its own Chatbot called LaMDA which recently a Google employee even claimed was sentient, but it has not been released to the public as the search giant fears how the public will react to it. Many people in the know believe that LaMDA could be significantly superior to ChatGPT which could bring negative attention towards Google as it is a poster child for what BigTech represents. And that’s why these technologies will probably become ubiquitous by the end of the decade save for regulatory roadblocks across countries.
Sahil Mohan Gupta is a Delhi-based technology journalist. He tweets @DigitallyBones. Views are personal.
(Edited by Tarannum Khan)