The Good A.I. We’re Not Talking About
A.I. has been out in the world for decades. With some very successful use cases. Going beyond the hype to the reality of A.I.
The past year’s headlines on Artificial Intelligence (AI) have mostly focused on one particular subset of AI, Large Language Models (LLMs) and Generative AI such as ChatGPT, Claude or Midjourney. For most of the general public that’s all AI is. Modern AI has been around for over 50 years. Algorithms, a core part of AI however, have been around for thousands of years.
As I’ve written before, AI is an umbrella term, a marketing catch-all that most scientists in the field of AI dislike. The hype around Generative AI (GAI), may even end up hurting the many advances and excellent, proven uses of AI across other industries.
Over the past year and a bit, as GAI slammed into society likes a bull in a china shop, it may seem as if we’re either on the path to techtopia or everything everywhere will end all at once. Neither is the case. Other AI tools have been around longer and proven themselves to be quite useful to humanity and are getting better.
It’s worth understanding and being aware of these other forms of AI puttering about in our sociocultural systems because, arguably, they are actually having and have had, more significant impact than we realize. Both good and bad, but then that’s the nature of how humans use technologies.
The Proven Artificial Intelligence Uses
I spent several years playing about with Natural Language Processing (NLP) a decade ago and have used LLMs a fair bit over the past year and been exposed to other uses such as Machine Learning (ML), Expert Systems and Deep Learning.
Through my netnographic research for technology companies, I’ve come to see how most of the general public misunderstands and tends to view AI as a whole.
Beyond Generative AI (GAI) there have been some interesting and very beneficial uses of AI tools in business and healthcare along with manufacturing, aerospace and more. Usually it’s just one type of AI that is used, sometimes a combination of two or three, but that is less of the cases.
Machine Learning
Perhaps the most common across a number of industries. Most of the general public would be familiar with ML, but not necessarily the term itself. Our most common engagement with ML is when it’s used for product recommendations and in social media feeds. It has also been helpful in disease diagnostics, fraud and spam detection along with supply chain management.
Deep Learning
This is a subset of ML (see how confusing it can get?) and applies Neural Networks (another subset of AI) in multiple layers to learn from massive volumes of data and often times, complex data sets. It has been used to analyse medical images, materials analysis in manufacturing and facial recognition as well as the development of autonomous vehicles and robots.
Natural Language Processing
If you’ve ever used Google Translate, you’ve used NLP. It’s also been used to help summarize complex documents such as scientific research papers and the creation of early and some current, chatbots.
Expert Systems
There’ve been a few cases of lawyers using AI to cite cases that turned out to be fake. They were using LLMs, not expert systems. Good expert systems used by law firms are trained on real legal data and have been used for several years quite successfully. They’ve also been used to support doctors, design products and help in engineering fields.
This is by no means an exhaustive list, but should help us better understand, beyond the hype of GAI, that the various tools that fall under the AI umbrella have been out in the world doing some very good things. Some bad things too, but that’s the case with all technologies.
These types of AI tend to simmer under the surface of the general public conversation. Mostly because they’re highly specialized and to most, not really that interesting in every day life.
The Realities of AI In The World
When we look beyond the hype of GAI, we can see that AI tools have been in use for a while. Some with great success and others that have flopped. So far the use of AI tools has lead to more jobs. This follows what sociologist Max Weber posited over one hundred years ago that the more technology a society uses, the greater the division of labour.
This concept has held true and was observable by Weber because of the history of technology creating more jobs for hundreds of years beforehand. While GAI may seem all encompassing, it is not.
Businesses are starting to use GAI, but mostly in experimental ways. There is significant caution on the one hand because of legal issues such as copyright and on the other because LLMs are proving to have a few problems that are too high risk for industry right now.
Another aspect holding back AI tools overall, not just GAI, is that many companies are dealing with significant amounts of technology debt. Old systems that need replacing, but require huge capital investment and so are managed over longer periods of time. Secondly, a company that has really good data management is the exception, not the rule.
For now, AI tools remain narrow in use, being used in some really interesting and helpful ways, but GAI is at best, a band-aid solution for most companies and a bit of a hot potato. Where we will see significant advances in AI use are likely going to be in what to most folks will see as, rather boring ways. It’s often these seemingly uninteresting applications however, that prove to be most beneficial to societies.