If AI engineers have their way, chatbots could soon be a distraction, thanks to Transformers. Transformers is a new kid on the AI block that aims to control all kinds of robots.
The ambition to achieve this is miles away, but dreaming about it will make the digital world go gaga, at least for now. This article retraces what AI scientists are dreaming and targeting to do with the knowledge and skills it has gained and accumulated so far.
The technology powering ChatGPT can do so much that was not envisaged earlier, and it is now feeding a sense of optimism. What is coming next will render the chatbox an unnecessary distraction.
AI advancement is geared to catapult chatbots and rise higher and higher. AI technologies are often linked to computers that can talk, write, code and create pictures built by transformers. And “transformers” have far-reaching applications beyond what has been achieved.
Google researchers were the first to write a report in 2017 that signified what the “transformers” were capable of doing. The report indicated that the “transformers” were a sort of AI algorithm that enables computers to understand the entrails of any data. After sifting through those data, the computer can churn out its own data developed from data, which can be in the form of a voice, driving data, or the amino acids in a given protein.
The immense capability of the “transformers” made OpenAI launch ChatGPT two years ago and made it a feasible proposition in the first place. Many AI companies are increasing based on the potential power of the “transformers.”
The innovation has enabled WAYMO’s robotic Taxi to design new proteins. Nvidia is now the MVP, and data centres sprawl worldwide, increasing electricity demand and bills.
READ: Distribution is King, Even in AI products, The Real Winners Might be Meta, Google and Apple
AI is excellent at figuring out patterns in information. Without transformers, AI abilities are very much constrained. Minus the transformers, AI processing capabilities are frustratingly very slow and limited.
With modern AI, assimilation and digestion of data are apportioned weight in a manner that translates meaning-based rather than literal-oriented. All this was unthinkable without the transformers.
The evolution associated with the transformers was to provide meaningful linkages between words and discern the intended meaning without being hazed by words.
Transformers and anticipation of the next sentence powered contextualisation. This feat made the generation of new information possible. That latent capability is now within reach to be applied in non-word data that includes images. The implications are simply breathtaking!
Some AI companies powered by the transformers are now vacuuming sequencing data from natural proteins and using it without the help of human scientists to learn how those proteins develop their molecular blockchains and how the protein it creates functions in the world. Learning has been simplified to a greater degree.
Once that knowledge is acquired, it aids in the creation of artificial proteins that can be deployed to resolve many human challenges, such as digesting non-biodegradable products such as plastics, and AI companies are in a race to develop universal knowledge that will subdue most industries to the knowledge they have
developed from the transformers powered AI. Some of their critics are beginning to label such imperial inclinations as “controlling freaks”.
New drugs that can treat diseases that are considered untreatable will be developed through transformers-powered AI. Stubborn tumours that refuse to shrink even when exposed to persistent chemotherapy may, shortly, meet their match in transformers-aided AI drug inventions.
Understanding snake poison and how it works, and either neutralising it or directing it to curative methodologies, will become urgent priorities because they are easier to learn than previously. Insects have survival means that are least understood yet can arm the human race with similar or better skills that are least appreciated today.
For example, ants can survive hostile climatic hazards by burying themselves beneath a mound and growing their own succulent food to sustain tens of millions and millions of its species. Humanity may have a lot to borrow and adapt to end global hunger.
In mining, transformers-enabled AI is now being used to scour what was once perceived as dead copper mines in neighbouring Zambia to locate unknown rich veins that can justify future restorations of abandoned mines to further their extractions.
Transformers-based AI will help scan geological features to determine viable mines, natural gas, fossil oils and gemstones without sinking millions of dollars in exploratory trials.
Conventional exploratory trials often ape a scattergun that misses more than it hits. Transformers-guided AI will reduce exploration durations, making research more predictable and lowering investigative costs.
Large language models used by ChatGPT, when combined with algorithms that incorporate images, have begun to yield sound results, such as robots that can perform human movements or tasks that do not necessarily require orderliness.
The profundity of this breakthrough means we are now in a stellar position where multifaceted thinking robots are within striking distance. While exciting, the drooling proposition presents credible new threats to humanity’s existence. Suppose the super-intelligent robots decide they no longer need us.
What will happen? Will we have the means to disarm them before the new contraptions send all of us to an earlier grave? It is imperative to note robots’ role in military conflicts is gaining new ground with untold implications staring at our frightened faces.
The capabilities of these robots will grow tremendously since we know the misallocations of resources to prioritize defence contractors and fund their confounding military conflicts.
Pre-transformer AI had besieged us with many headaches. The software-powered robots had, at times, struggled to cope with challenges given their limited data reservoirs and digestive and understanding bottlenecks, condemning the users to an infuriating crossroads.
All that is about to end is gratitude to transformers-aided AI that can chew “visual language models” where language and images are interconnected. Current technologies have been floundering because data has not yet been linked with images.
Once this deficiency is ousted, we should expect driverless machines to perform perfectly that human drivers can only envy and marvel.
Transformers-aided AI will be able to identify obstructions like an imposing broken vehicle in front and assess alternatives before deciding what next step to pursue and outdo the hindrance.
It could be to turn back or reverse or sidestepping the broken car. Paradoxically, current AI driverless vehicles are bamboozled to address such minor miasma.
Human labour replacement is long in the making because such powerful and intimidating technologies still carry their own risks, reflected in their unpredictability, unreliability, and overdependence on humanity to perfect them.
They also face formidable limitations imposed by the available data and their capabilities to translate it into useful information.
In certain repetitive tasks, transformers-assisted AI will take over because most data is available, and there are few unpleasant surprises to contend with.
Data collection and feeding to transformer-aided AI to replace human labour may be too costly in certain fields, preserving the current status quo to the relief and delight of many who depend on those jobs to make a living.
Whether we like it or not, transformer-aided AI is here to stay and will get even more powerful, upending human lifestyles in a manner we can hardly imagine.
