AI and its ever-expanding skillset
Reading Time: 3 minutes
Is There Anything AI Can’t Do?
With artificial intelligence, the possibilities are endless.
These machines, models, and algorithms become more human every day – they basically squeeze the decision and discernment power of an active human brain into codes. The results already go beyond our limited capacity. Thus, that AI as the future of work and productivity is an understatement; it should be regarded as THE future.
Artificial intelligence can now prescribe medicine, and experts believe this will positively impact overall health.
Research shows medication-related errors are responsible for one out of 131 outpatient and one out of 854 inpatient deaths in the United States. The act of prescribing medicine is as complicated as it is vital, so medical experts must practice for years before gaining mastery. AI, however, is on the verge of replicating their skills and making prescribing as easy as clicking a button.
This is yet another example of artificial intelligence beating humans at their own game. For instance, autonomous AI models can already identify breast cancer better than teams of certified oncologists. It has also been recently reported that AI can successfully discern COVID-19 from cough sounds.
For now, let’s examine how AI models can prescribe medicine.
GPT-3: So Good It’s Scary, Literally
The internet has gone crazy for GPT-3 (Generative Pre-Training Transformer-3) from OpenAI. It is a third-generation machine learning model that designs websites, answers questions, and yes, prescribes medicine.
OpenAI has set the AI-world on fire with debates about safety before; the introduction of GPT-2 was met similarly.
Although GPT-2 boasted many benefits, critics termed it too dangerous because it could create writings that were indistinguishable from those penned by actual humans. It was TOO good.
At the time it was considered to be so dangerous, GPT-2 was only utilizing 124 million parameters of the possible 1.5 billion in its original design. Well, watch out. GPT-3 will feature a staggering 175 billion parameters.
GPT-3 is a neural-network-powered language model that predicts the likelihood of a sentence existing in the world. It leverages a generative model of languages (two neutral networks perfect each other by competition) and can acquire knowledge and process long-range dependencies by pre-training on diverse sets of written material with long stretches of contiguous text.
As a language processing model with the largest database of training sets ever at its disposal, experts believe it can answer medical-related questions, diagnose asthma, and prescribe medicine.
Google’s Prescribing Model
As powerful as GPT-3 is, it is not the first AI-based model capable of prescribing medicine. Google’s AI could predict the scripts a physician would write with up to 75% accuracy.
The most significant setback of Google’s AI is that it is rooted in historical data. It can only replicate the physician’s prescription patterns and not their ever-expanding knowledge of medication and side effects.
Based on the researchers’ submission however, the system, if appropriately applied to healthcare, could assist physicians in identifying abnormal or incorrect prescriptions. It would work similarly to the fraud detection programs that banks employ.
1. Artificial intelligence is our future – that becomes clearer and clearer every day.
2. AI can reduce the human error associated with prescriptions.
3. OpenAI and Google are well on their way to providing AI solutions that will correctly prescribe medicine.
Comment and share! For more content related to technology and medicine, check out AIM Blog.
- Stanford’s AIMI is Revolutionizing Healthcare AI by Providing Free Big Data to Researchers
- Using Video Games, Gametherapy Is Making Healthcare More Fun and Engaging
- IoT-Aided Robotics and Healthcare’s New Revolution
- Doctors Are Leaders, but Are They Good at It?
- The Relationship Between Stress and Gut Microbiota