We remember those years where a seemingly minor event led to a chain reaction that changed the world such as 1698 when Thomas Savery produced the first practical industrial steam engine or the article in Nature in 1953 by Crick and Watson describing DNA as a double-stranded molecule that could both produce exact copies of itself and carry genetic instructions. No doubt, the burgeoning field of bioengineering 70 years later will only amplify this seminal date.
It may well be that 2022 will be remembered not as the year that Russians invaded Ukraine and changed the post War geopolitical order but as the year that Artificial Intelligence was finally delivered for general usage. Artificial Intelligence has remained in the world of software research and academic discourse for decades but on 30th November 2022, OpenAI released ChatGPT. It received 1 million users in 5 days, 100 million users in January and currently has more than 13 million visitors daily. No software has ever been adopted so fast and it remains absolutely sensational!
We wrote an internal memo to understand the phenomena and if you need a basic understanding of ChatGPT, we have published it for you to read here.
We said of it in December “it is a harbinger of the AI that we have been promised for years but has never been delivered. In a nutshell, it is able to intelligently replace a google search by finding all the associated data and then formulating an answer that is equivalent to a university student’s thoughtful reply. It can also produce programming code at a credible level of accuracy. To get a log-in to beta mode AI click https://chat.openai.com/chat”
Subsequently, Google, who achieves the vast majority of its revenue from Google search advertising, has announced its Bard AI system (currently available in the US & UK only) and Microsoft, a major investor in Open AI, has built GPT-4 into its Edge browser via the Bing search engine. Confirmed: the new Bing runs on OpenAI’s GPT-4. Another useful link is (The new Bing – Learn more). These announcements suggest that very shortly we will be doing internet searches that provide an answer rather than a list of websites and internet documents to trawl through. The substantial issue currently is knowing whether AI’s convincing answer is correct.
Nevertheless, December 2022 can be described as a “Netscape Moment” (the 1994 launch of the web browser that made it easy for ordinary people to surf the Internet) and it clearly is a “Code Red” threat to Google who have dominated the Internet Search world for 20 years.
In the 1990s economic historians started talking about “general-purpose technologies” (GPTs) as key factors driving long-term productivity growth. Defining attributes of these GPTs were held to include rapid improvement in the core technology, broad applicability across sectors and “spillover” (the stimulation of new innovations in associated products and services). Think printing presses, steam and internal combustion engines, electric motors and transistors. Last December’s announcement positions AI as a GPT.
By the way the acronym GPT for General Purpose Technology has no relation to Open AI’s use of “GPT” which stands for Generative Pre-trained Transformer although it could become a neat double-entendre!
Artificial Intelligence can broadly be described as software that simulates human behaviour or thinking and can be trained to solve specific problems. There are numerous competing technologies and rapid progress will be enhanced by the ability of the leading contenders’ software not only to be trained but to “self learn”. It should be understood that ChatGPT/GPT-4, and other similar conversational AI systems, only represent a sliver of AI that relates to digesting and making sense of vast volumes of data in conversational mode. However, it must be added that ChatGPT/GPT4 also has the capability of producing computer code drawn from banks of sample code found on the internet, creating haunting images in specified styles and authoring fictional stories. As such this early AI version must be seen as both astonishingly powerful but also demonstrating the power of AI that will, in due course, be applied to wide areas of human endeavour. We can now expect the imminent use of AI in any area where vast stores of data or billions of parameters or rules apply. Examples include complex pattern recognition such as in blood cell analysis, legal assessment, medical diagnosis, plant and animal genetic research and editing, robotics & humanoid development, risk assessment and decision making, face & speech recognition, autonomous driving (in its broadest sense) and personalised marketing to name a few. On the military front the possibilities are terrifying and we already see loitering drones in the air over Ukraine equipped with automatic target acquisition algorithms.
In time, specialised chips will dramatically enhance AI performance; billions of documents or images with billions of parameters will be able to be instantaneously accessed and digested and an information age, as we cannot imagine it, will be upon us.
PWC estimates that Artificial intelligence (AI) can transform the productivity and GDP potential of the global economy and that 45% of total economic gains by 2030 will come from AI led increases in product variety, personalisation and affordability. AI algorithms will enable personalisation and tailor made products at a level unknown in history. PWC suggests the potential global impact could be in the region of $15.7 trillion by 2030 and OpenAI research predicts that 80% of U.S. workers’ jobs will be impacted by GPT.
Last year American venture capitalists invested a record $115bn in AI companies, according to PitchBook, a data provider and China is making the field a national priority. Hot competition between China and the USA almost guarantees the rapid integration of AI software into competitive products and services.
Mantis Networks has been interested in AI for some time and it was one of the reasons we entered the CRM world as we knew that it would lead us into Expert Systems (workflow management where much of the revenue is) and subsequently to Artificial Intelligence. AI has already made its first small appearance in Bitrix24, the CRM that Mantis distributes, in the form of sales data analysis that shows the probability of winning a lead and which events influence the probability of winning leads the most. Its AI functionality ranks potential customers according to predetermined criteria and the likely value of each customer. This way users can prioritise their customer interactions and select the most effective steps to convert the highest probability leads to sales.
Imagine the potential to improve worker productivity when a later generation of AI system is “trained” to distil meaning from an organisation’s internal corporate data sets such as prospects and orders, procedures, product manuals & specifications, contracts, books of account and correspondence etc.
We are both excited and deeply worried by the potential of this now, real General Purpose Technology. We also ask where South Africa and other developing economies are in this freshly defined “information age”.
Mantis Networks will try to stay abreast of developments so expect more from us on this subject shortly.
However, what we can say at this stage is that we are planning to research how and which AI iteration can best be used to enhance our product offering.