The Future of Generative AI
The impact of AI on the future of mankind.
For many weeks I’ve been resisting writing an article on the future of AI because I don’t feel like I can really do it justice. So many people talk about the future of AI citing the more recent technology that’s being produced as well as possible applications. I’ve read several papers, looked into the underlying technology, and reviewed how people are quickly adopting this into a suitable business tool that can start, level up, or even pivot their idea. But in my opinion the majority of these articles are missing the fundamental point because they’re investigating 1–2 year time horizons. What we’re really seeing is the evolution of our society as a whole.
Learning will be redefined
Time travel back a century and information was communicated by books: the mind poured into paper, massaged into text, and if you’re lucky a professor with ample knowledge of the subject could likely help stream the information into your brain after a bit of forging and hammering in lecture and discussions.
Then came the internet and the age of video where information was seemingly at our fingertips, almost everything was there in the worldwide web if you had the patience to find it properly but again we were fundamentally digesting information in it’s unprocessed form: sure the blog is easier to read or the lecture easier to watch but the information did not change: if something didn’t make sense you’d likely have to search again, read another book, and this takes time.
Enter Generative AI, where information is retrieved based on how it is requested: it’s like picking up a book where you get to write the title and after writing the title the whole book is filled with exactly what you were looking for. If a chapter didn’t make sense then you could ask for a supplementary chapter to help guide you. These technologies are making information digestible to every format, i.e. you. The accuracy of the answers, the quality of the response, and even the modality of information (text, audio, images) will all get better. Soon experts and concepts won’t be human: they’ll be algorithmic representations that have been verified by experts. Biology concepts, psychology concepts, physics concepts will all be contained in a model that can teach these concepts at levels spanning from a 5 year old to an expert in the field. In a few years, traditional learning will be (and should be) a thing of the past and the teachers will just be there to have discussions that validate information and fill in the gaps. I wouldn’t be surprised if in 20 years we’ll have holographic-like tutorial videos that can walk an individual step by step through a task or learning objective, like improve a golf swing or soccer kick, building something like a table, or analyzing something like the human brain.
Accessibility
Even more powerful is the accessibility of this technology. There’s a high correlation of inherited expertise and societal value; unequal access to opportunities disproportionally effect someone’s ability to grow: Elon Musk was the father of a significant electromechanical engineer, Bill Gates had early access to computers at his private school, Steve jobs in elementary school was almost kicked out because he was bored but fortunately he had a teacher who motivated him. Again, this is not a necessary condition and there are many self-made millionaires and billionaires, successful authors like JK Rowling, etc but the access is still an unfair advantage. Generative AI has the potential to make information accessible to all and while this will not improve the network effect this can improve early education and possibly jumpstart the minds of many brilliant individuals who had potential but not the same opportunities.
So where does this leave humankind? Objectively, these models are creative and even in their nascence they display interesting emergent properties but for the time being this technology and its outputs are fundamentally used to empower people. We’re just changing from builders to farmers. Soon we’ll focus less on coming up with an idea and focus more on farming for ideas and combining them with our thoughts. Want to learn how to build a PC from scratch, how do you start? Want to learn how to start a company, how do I begin? Want to learn how to draw, can you teach me how? Would this schemata work? Given my journal notes, what makes me happy? How do I invest in stock? How do we get to space? How do we solve climate change? How do we make education accessible to the world? The limit’s our imagination, the questions we can write, and the speed at which traditional engines adopt these technologies (like government, academia, etc.). Soon even our questions won’t be text, we’ll be able to ask in pictures, sounds, and videos.
Tentative Predictions
I predict that within the next 10–20 years the following will happen:
- Significant advancements in technologies (particularly in hardware and software industries) due to having Generative AI in the development loop to design architectures, models, and training strategies
- Company brands being represented as specialized generative AI models that are tuned to the images, videos, text, slogans of the company.
- Significant changes in education and an adoption of generative AI technology in school that accelerates individual learning.
- Significant improvements to the healthcare system as large corpuses of medical data can now be synthesized in a matter of moments.
- Improved individual well-being as access to this technology empowers individual growth in all interests.
- Possible AI sentience that will fundamentally reshape the way we perceive consciousness.
- Automation of many different industries and jobs: driving, secretary, second brain.
- Emergence of these technologies as a platform from big industry companies: google, microsoft, IBM, oracle, etc. for personal and professional use.
Curious?
Haven’t tried it out ? Take a look (this is just an intro list and is by no means exhaustive)
Text-to-text
- Open AI’s ChatGPT3: https://chat.openai.com/
- Microsoft’s Bing search engine: https://www.bing.com/
- Google’s BARD: https://blog.google/technology/ai/bard-google-ai-search-updates/
Text-to-image
- Midjourney-v5: (https://midjourney.com/ you’ll have to integrate it on a discord server first).
Cool applications
