By puttin' generative AI functions inside programs, we fit create ways wey no fit happen before with normal programming.
Apart from dat, as generative AI go fit automatically generate programs, we go fit freely and easily create and run programs as we tink dem.
I don build systems before wey dey translate my blog articles to English and post dem to an English blog, create explanatory videos from presentation videos and upload dem to YouTube, and generate and publish my own blog site with indexes, categories, and tags.
For dis way, one system wey dey use original content as raw material and dey put generative AI functions inside to produce different different kyn content fit be called an Intellectual Factory.
Wetin pass dat, I don create one web application to run dis Intellectual Factory and manage how e dey work, wey go make am accessible for both PCs and smartphones. Also, di parts wey dey handle automatic processing wey events dey trigger, dey run on virtual machines wey dem prepare for batch processing wey pass di backend.
So, I developed di PC and smartphone frontends, di web server backend, batch processing on virtual machines, and di infrastructure for dem, all by myself with di help of generative AI.
Dis one no just be full-stack engineering, but e fit be called Omnidirectional Engineering, wey dey completely develop different different parts of a system.
Apart from dat, when I wan improve wetin no dey convenient for di developed web application or add new features, I fit give di programming work to generative AI, wey go make am easy to improve as I dey use am.
Dis one dey even more flexible and fluid pass normal software, e dey allow me to create something wey fit my usage perfectly. I dey call dis one Liquidware.
I don personally develop and I dey actually use all dis ones. Dis one no be just idea; e don already be di reality of software development.
Even though dem never develop am finish, for di area of business systems, I dey expect say di Business Process-Oriented development method go become real.
Dis one na one way wey no dey aim for general optimization of programs, wey dey make systems complicated, but instead, e dey divide software modules into individual Business Processes.
Only di basic framework definition of di user interface, how to manage user privilege, and data models wey suppose dey shared between Business Processes, na dem go share as di outer framework of di business system.
Other internal system processing and temporary data dey managed as part of di Business Process.
E fit get functions or data structures inside dem wey two or more Business Processes fit share. But if dem turn dem into shared modules or custom libraries, even though code and quality reusability go improve, di software structure go become complicated, and if dem change am, dem go always need to consider how e go affect other Business Processes.
For a situation where generative AI dey automatically generate programs, di disadvantages of di latter pass di advantages of di former. Because of dis, di Business Process-Oriented way, wey dey emphasize Individual Optimization instead of general optimization, go make sense.
Wetin pass dat, imagine things like "entering new employee basic information," "updating employee basic information," or "searching for employees by name" as individual Business Processes.
For old-school development methods, each user interface, frontend process, backend process, and batch process go dey separated into different files for different directories. Also, different engineers go develop each one.
But, when one single engineer dey do Omnidirectional Engineering by making generative AI do di programming, e make more sense to put all di code wey dem need for one Business Process inside one single file or folder.
Wetin pass dat, results of requirements analysis, test specifications, test results, and review records fit also dey put togeda for di same place.
Dis one go make am possible to manage all software engineering artifacts as one single Business Process unit. And, because no need to think about general optimization, dem fit focus on improvements inside dat Business Process, and dem fit easily add new Business Processes to di business system.
For dis way, how dem dey develop programs and wetin programs fit develop dey change well-well because of generative AI. Dis one no be for future; e don already dey happen now, and for near future, di level of perfection go just continue to dey improve, and di next stage must surely pass dat.
Simulation Systems
Wetin programs fit do no just stop at di business systems and intellectual factories wey I mention here.
Di other areas wey I no mention fit be broadly called simulation systems.
Whether na to solve simple physics equations with one single analytical formula or to calculate complex physical phenomena with programs wey dey repeat, both fit be called simulation systems.
Wetin pass dat, dem fit use simulation systems no just for physics, but also for chemistry, biology, sociology, economics, and other fields. Apart from school work, dem dey also use simulations for engineering, medicine, how institutions dey run, and business management.
Games self na one kind simulation system. For any game, di physics, society, rules, and other things inside dat game's World, na like dem dey simulate am.
Wetin pass dat, we dey also do one kind simulation when we dey plan our lives, travel, or how to spend our pocket money.
Dem don do dis simulations for different ways: by creating and running programs, calculating equations on paper, thinking for our heads, arranging thoughts with text and arrows on a whiteboard, or drawing graphs for Excel.
If dem develop a simulation program for one specific problem, e go allow for more complex simulations pass analytical equations. But, e go need programming development skills, effort, and time.
E also need to make di simulation model clear, wey in turn go demand skills, effort to think, and time.
Wetin pass dat, simulations don dey limited to wetin dem fit express for program form, and before, only wetin dem fit express computationally fit be simulated.
Generative AI dey change dis situation well-well.
Generative AI no just fit easily develop simulation system programs, but by puttin' generative AI inside simulation systems, dem fit also simulate things wey no fit be expressed mathematically. Dis one dey make ambiguous qualitative simulation elements and simulations wey involve human-like intelligent agents possible.
Wetin pass dat, dem fit express dis simulation models no just mathematically but also for natural language and generative AI fit interpret dem.
Dis one go make am easy to change di different simulations wey we don do for many situations into simulation systems.
As a result, we go fit get more accurate, efficient, and effective simulation results, with small possibility of missing details or putting biases.
Wetin pass dat, when we dey consider or discuss complex problems, we go fit use a simulation system for consideration and discussion, instead of just thinking for our heads alone.
Dis one dey make consideration more precise and make discussions more constructive. Instead of pointing out each other's intelligence or mistakes for thinking, discussions fit focus on clear points like di models wey di simulation dey based on, any omissions or missing elements, how dem dey estimate parts wey dey very uncertain, and which metrics among di results dem dey prioritize.
As simulation systems dey become easy to create, our way of thinking go change from linear thinking—wey dey focus on intuition, assumptions, and di bad intentions or mistakes of others—to Simulation Thinking.
Dis one be like searching di internet on a smartphone during a discussion to verify news sources, Wikipedia, or primary sources. No go need for endless arguments wey dey rely solely on each other's memories.
During a discussion, generative AI go arrange di simulation model, simulation rules, and preconditions from di content of di discussion.
Participants for di discussion only need to add or correct information and premises for dat model and rules, and den check di simulation results. Just as when dem find a credible news source, dis simulation results fit serve as common ground to deepen di discussion.
Dis one mean say pipo wey dey listen to di discussion no go longer live for one era where dem need to ponder who is right or who is trustworthy. Dem no go also lose sight of di main point by trying to understand difficult technical terms and concepts wey dey appear for di discussion.
Dem go only need to consider very simple things: how to evaluate uncertainty and which values to prioritize.