The monkey business of Neuralink
The Neuralink breakthrough and its impact on AI
Elon Musk is like no other. He is the billionaire with an appetite for the impossible yet he stands on principals, first principals.
As he says, “Going back to the fundamental truths and working your way back”. This approach has helped him conjure up many ideas which led to companies which led to success stories.
Along with Tesla, SpaceX, OpenAI, Hyperloop and more companies, Musk has another venture that has been submerged in secrecy. Neuralink is, as their website indicates, a company that is ‘creating the future of brain interfaces building devices now that will help people with paralysis and inventing new technologies that will expand our abilities, our community, and our world’
The company has been more open about its projects recently with a demonstration led by Elon Musk in August 2020 and most recently, in April 2021.
The latest stunt involved “Pager”, a macaque, who was put in the pilot’s seat of Musk’s newest exposé. The experiment was quite simple, Pager would use a joystick to play a brain stimulating game of ~ follow the square ~ as Darrell Etherintgon from Tech Crunch describes ‘it [Pager] had to move a token to different squares using a joystick with its hand’. This type of pre-experiment is called a baseline where Neuralink used sophisticated algorithms to learn Pager’s move and anticipate them before they occur.
Pushing the envelope even further, Pager was presented with a game of Pong, bouncing a ball side to side with a joystick. For the following experiment, the joystick was removed yet Pager had its eyes glued to the screen. Lo and behold, the embedded neural threads helped Pager guide the paddles as if he had his joy sticks on. This was made possible thanks to the sophisticated Machine Learning (ML) models.
What happened is that the system captured electrical signals emanating from the cerebral cortex, the part of the brain that coordinates hand and arm movements, thereby designing a pattern between those neural signals and the different gestures made by the joystick.
Pager is seen bouncing the pong ball from side to side with now joystick
This is a remarkable achievement from Neuralink. Musk has made bold claims even suggesting that improved iterations will be able to help paraplegics regain their motor and/or sensory functions. Hence, breakthroughs in science and technology are always welcomed with enthusiasm, sprinkling a dash of hope to unanswered questions. Yet Neuralink has high hopes and AI is at the heart of it all.
AI seeped through the cracks of the untouchable and incomprehensible. Today, AI is more accessible, organizations are seeing the benefits it could bring them. No longer subjects of use cases, AI applications are solving real problems.
The technology is not always stand alone; it is basically incorporated in areas where prior methods have failed to fill the gap. For instance, scientists understood that they had to mimic the neural signals stemming from the limbs for the 2,000 implanted electrodes to push the brain to respond. But getting that accurate representation of the responses required analyzing a plethora of data points. This where Machine Learning comes in, a subset of AI, building models to correlate neural activity patterns with hand movements.
Most of the mundane will now have some sort of an AI component to it, and the mundane consists of our jobs and what we use to do them. For companies using applications like enterprise software, getting the most out of their systems will boost their ROI. Enterprise grade AI software is relatively new and most providers struggle to GTM (Go-To-Market).
It’s usually a daunting task to get an AI product on the ground but a lot of implementations are based on potential use cases. One use case is the Instance Tracker: the user will be monitored in different scenarios through the different tasks they accomplish (the buttons they click, the inputs they encode, the amount of time they stay on a specific page, so on). By analyzing these data points, a ML algorithm can help the system identify efficient use of the user’s time and effort by recommending the next ‘optimal action’.
Another use case that is applicable is in the manufacturing industry where the deployment of IoT devices has become main stream. IoT stands for Internet of Things; they are basically devices that are interconnected via a network with the ability to transfer data over the internet. IoTs are proving to be great in reading and relaying information to the controller. Data engineers are leveraging ML to predict when and how different machinery will fail. Mechanics can layout a strategic maintenance plans to stay ahead of the issues.
That being said, ML models can go beyond that especially when autonomous robotics is being used more and more in the industry. These robots can now be fitted with IoT devices and simultaneously predict the failure instances of the machinery and conduct the appropriate fixes. This approach goes beyond learning; it is built to trigger a response, just like Pager’s neural activities.
Industries like car manufacturing will benefit immensely in deploying AI powered autonomous robots
Imagine a human operator standing by machinery just looking at the activity board for any impending failures. He sees stress points elevating, the IoT sensors sense the danger too and alert the operator to take a list of suggested actions. His response is monitored; every gesture of the attached robotic arm he makes are recorded and catalogued.
But what if these co-ordinations were skills that were acquired? Machines will benefit from the immense experience (‘data’) in learning from the human operator. They will then be on standby, strapped with their IoT devices; they will recognize the situation and mimic the human robotic gesture responses. Experience will help them re-think their approach midway if a new circumstance presents itself.
To sum things up, Neuralink’s quantum leap further extends the possibility of using AI to curb some of the world’s biggest problems. Elon Musk is adamant, even begging people to apply to his company that the long term applications of these technologies are as he tweeted “species-level important”.
AI will be common in our lives, it will change the way we think, the way we live and the way we work. Its reach will widen in most cases and strengthen in others. Enterprise and Software will still go hand in hand, but this time, AI will bridge the gap and get us from ascertaining patterns from data to letting machines make their own assessments thereby giving them a brain of their own.