TO AI OR NOT TO AI (January 2025)

A quick crash course on AI.

Reactive Machines. These are the most basic forms of AI that respond to inputs with pre-programmed actions, like chatbots on websites. I don’t like to call them AI because they really have no intelligence at all and have absolutely no ability to actually learn or self-write memory. They are only good for what they were programmed to do.

Next, there are Limited Memory machines. This type of AI can use past experiences and data to inform future decisions. In other words, they have the ability to learn from the past and not repeat the same mistakes. An example of this type of AI can be found in self-driving cars. While technically not perfectly correct, Artificial Narrow Intelligence falls within this family of AI—in terms of intelligence at least, but it’s a different type. These family members include AI like ChatGPT, which is a Natural Language Processing AI, and cousins like Deep Learning, Generative Models, Pre-trained AI, and Contextual Understanding AI. Also in the ‘family Christmas photo’ and part of one or more of these categories (because there is almost always overlap) are Amazon’s Alexa, Apple’s Siri, Netflix, and Spotify, and even AI apps for healthcare. These are all at the top of their game, and the next category above these does not yet exist.

We must remember AI is not just software but in many cases has a physical body or form, such as robots and humanoids: Spot, Atlas, Sophia, ASIMO, Tesla’s Autonomous Driving AI, and Pepper by SoftBank Robotics, are all prime examples, or our ‘Optimus Primes’ of 2024.

The next level includes Artificial General Intelligence and Artificial Superintelligence. Remember, none of these yet exist. Within this family extension are Theory of Mind AIs. All these AIs are in experimental or theoretical stages. These AIs would have some, if not all, of these characteristics:

• Human-Like Understanding
• Complex Problem Solving
• Autonomous Learning

This family of AI would be comparable to a human’s entire brain and nervous system, whereas other AI (like the ones we have today) are basically single-purpose living cells.

Superintelligent AI will surpass human cognitive ability, be able to self-improve and ‘self-replicate’, have emotional and social intelligence, contextual awareness, and, I hope, ethical and social insight. In my lifetime, I will see this: a robot with AI that can ‘e-breed’ or replicate, and with each new creation, make an improved version of itself. An interesting nugget to know is that ‘cell to chip’, our brains are still the most energy-efficient complex processing machines. The only other machines that can emulate a similar number of calculations per second use entire floors of PCs needing huge amounts of power. This means that before we can ever get access to ‘AI at scale’ or ‘AI of full potential’, we must solve this power problem. As of the time I prepared my keynote talk, Microsoft had just announced plans to take over a nuclear power plant to power its AI data center ambitions! Google will also be building nuclear power plants.

Anyway, I digressed a little. AI will be made in our image, and it will have surpassed us in terms of intelligence within my lifetime. I estimate this to take place between 2030 and 2045, though I keep updating this prediction. But, well before this date arrives, business and society will have fully integrated AI into just about all parts of work and play. The truth is, AI is really already here. It’s just on a leash.

Let’s look into the future now, with a summary 5-year global scenario.

I believe AI is expected to become deeply integrated into our everyday homes and offices. The major changes will be increased autonomous ability and much higher intelligence, making ChatGPT-4 look like a teenager and humans like infants. What will an AI like this look like? Hopefully invisible! I predict AI in the future will be largely invisible, and this will be by design. People won’t really want to hear or see AI unless called upon, and a lot of work will be done to ensure AI does not dominate our space.

Unfortunately, there is an area where AI is already a tad out of control: the realm of marketing. In marketing, AI is already being both abused and abusing! While doing some research, I came across an article about an AI bot clicking on paid adverts to generate income.

I believe the biggest positive advancement of AI will probably be in our homes, in the forms of wearables and interfaces. I don’t know what device is going to replace the smartphone yet, but something will. The device will be AI-driven and probably integrated in some way into our bodies. I predict we will see this in 10-15 years. Our new AI counterparts and AI assistants will often have virtual similarities to ourselves or at least adopt some of our personality traits so we can get along better with them.

At the office, work, or factory, I believe AI will be more obvious and intrusive. I expect it to have more physical presence. It would need to because it’s not just crunching data and voice commands; it will need to perform dangerous and physically challenging tasks. Regardless of whether it’s home or office, for those who are willing to embrace AI, we will be part of the AI-Collaboration era. This will be a period of time spanning a decade, with sometimes awkward and sometimes successful experiences.

AI will play the part of the complex decision-maker in the team wherever lots of data and data sources are at play, such as in electricity generation and distribution, water management, healthcare, and law and justice, to name just a few.

The first real foundations of smart cities (where AI will need to ‘live’) will arrive a little after this, between the years 2035 and 2045. Only once we have ‘smart spaces’, and have resolved the challenges of the ‘AI environment’ will AI reach its potential. I believe the first smart cities in Africa will be Cape Town and Kigali.

On the periphery, in the media and courts, we will see a great deal of legislation designed and opposed. There is a window of opportunity for businesses to become involved in the development of AI legislation, and the need for input is high. Unlike simpler technologies like the microwave or fax machine, AI is more comparable to electricity—it can be hazardous if misused, but it will power and drive all future technologies, presenting both challenges and immense potential.

The elephant in the room—or in the lab (because of who this keynote talk was for)—is job displacement. In a country where over half its youth are without jobs, removing potential jobs from the unemployed is a complex issue. South Africa faces a job crisis, among others, and we have to tread carefully regarding AI. It’s crucial to focus on future jobs and upskill the current workforce simultaneously.

For the client I was offering a talk to, it was important to share a few custom nuggets for them, and without giving away any privileged information, I want to share just a few items that could be applied to most companies in manufacturing.

Over the next five years, technology will greatly influence businesses like yours and others offering scientific and laboratory equipment, instrumentation, and consumables. You are in a niche sector of manufacturing and distribution vital for the pharmaceutical, chemical, and food and beverage industries.

AI can be used to enhance productivity and ensure accuracy. I have no doubt about that. However, efficiency can also come at the cost of human jobs, so when we implement technology, we must do it ethically. It’s a catch-22 situation if ever there was one. I do believe humans will suffer the net loss here, unfortunately.

Management teams need to imagine what a future laboratory might look like, including all its operational components. So let’s imagine:

Real-time data for more accurate analyses. This is literally a product ‘BOBSCOMPANY’ can and must offer. In many ways, I suspect your manufacturing floors and laboratories will become showrooms.

Digital transformation in the area of customer interactions is likely where technology and AI will see the most change in the short term—simple things like smart chatbots on your e-commerce platforms that handle real-time orders in multiple languages.

Technology can be used to build brand awareness and reach future clients in digital spaces, reinforcing your company’s image as a key provider of cutting-edge solutions.

I also envision your devices or solutions being integrated in real-time with IoT for remote monitoring and support; tech can be actioned on the road or on-site for client sites—this could be the future advantage you target before your competitors. An example of a company that did this well is We Buy Cars.

Some software’s which includes AI to keep an eye on, I’ll share these in some categories:

As I end this article, let me do so with some tips on where you might want to start if you do want to implement AI. In just about all businesses, you will have AI for machines and AI for humans.  If applicable, you might want to research and do some development around Predictive Maintenance. There are AI start-ups in South Africa you can collaborate with. Some examples I gave in my keynote talk were DataProphet and Laminin.

Embrace and investigate the circular and subscription economy. In the future, people and businesses will want to own less equipment, and the equipment they do have must be maintained and swapped out for newer versions effortlessly. This is something your sales team can design and attempt to perfect before competitors.

Approach AI with a healthy curiosity and a reasonable predetermined exploratory budget. Monitor competitors. Collaborate when a mutually beneficial opportunity is found. But most of all, instill a culture within your teams to investigate and spearhead implementation into their respective departments. Make decisions that serve the humans, in the company and out.

Remember, I offer digital scenario planning and I am also a digital architect. If you are keen to chat to me to learn more about how to implement AI into your company, you can reach me via my websites contact page.

The END.