By putting generative AI power inside programs, we fit make tin wey before before, normal programs no fit do.
Plus, as generative AI go fit de automatic make programs, we go fit make and run programs wey we tink of, as we like am, and e no go hard.
Te te now, I don make systems wey fit change my blog tori dem enter English and post am for English blogs, make explain videos from presentation videos and carry am go YouTube, and make my own blog sites wey get index, categories, and tags, come publish am.
Like dis, a system wey de use original content as raw material and add generative AI features to produce different different content wey come from am, we fit call am "intellectual factory".
I don still develop one web application to run dis intellectual factory and manage wetin e de do, so e go fit work for PC and phone. Apart from dat, di parts wey de do automatic work wen something happen, dem de run am for virtual machines wey dem prepare for plenty plenty work for di backend.
So, na me one person develop di PC and phone frontends, di web server backend, plenty plenty work for virtual machines, and di things wey dem need to run, all of am with help from generative AI.
Dis one no just be full-stack engineering; e fit be "omnidirectional engineering," because e involve developing different different parts of di system completely.
Furthermore, wen I want make di web application wey I don develop beta, or add new things, I fit give di programming work to generative AI, so e go de easy to make am beta as I de use am.
Dis one make di software even more flexible and free pass di old software dem, e make me fit make something wey go fit wetin I de use am for perfectly. I de call dis one "liquidware."
I don actually develop dem and I de use dem now. Dem no just be ideas, dem don turn to reality for software development.
Even if I neva develop am, I de expect say for business systems, di way dem de develop wey dem call "business process-oriented development" go become real.
Dis na one way wey no de aim to make programs work perfectly overall, wey de make systems hard, but instead, e de separate software parts based on individual business processes.
Na only di basic framework of di user interface, how to manage user permissions, and data models wey need to de shared between business processes, na dem dem de share as di outside framework of di business system.
Other inside system work and temporary data, na at di business process level dem de manage am.
Dis fit include functions and data structures wey two or more business processes fit share. But, if dem make dem shared modules or custom libraries, even if code and quality go better for reuse, di software structure go come hard, and changes go need you to de always tink about how e go affect other business processes.
For a situation where generative AI de automatically make programs, di disadvantage for di second one big pass di advantage for di first one. Na why di business process-oriented way, wey de focus on individual making-beta instead of overall making-beta, go make sense.
Still, imagine things like "entering new staff basic information," "updating staff basic information," and "searching for staff by name" as individual business processes.
For di old ways of development, dia user interfaces, frontend processes, backend processes, and batch processes, dem de separate dem enter different files for different folders. And again, different engineers de develop each one.
But, wen one engineer de do omnidirectional engineering and de allow generative AI do di programming, e make more sense to put all di code wey one business process need for one file or one folder.
Plus, requirements analysis results, test specifications, test results, and review records, dem fit also put dem all for di same place.
Dis one make all di things wey software engineering de produce fit de managed for each business process. And because you no need to de tink about overall making-beta, improvements fit focus inside dat business process, and dem fit easily add new business processes to di business system.
Like dis, how dem de develop programs and wetin dem fit develop with programs don change well well because of generative AI. Dis no be something for future; e don already de happen now, and for near future, e go just de better, and di next stage go just go beyond dat.
Simulation Systems
Wetin programs fit do no just stop for business systems and intellectual factories wey we talk about before.
Di other areas wey I neva mention, you fit classify dem generally as simulation systems.
Whether na to solve simple physical equation with just one analytical formula, or to calculate complex physical happenings with repeated programs, you fit call both of dem simulation systems.
Asides dat, dem fit use simulation systems no be only for physics, but for chemistry, biology, or even sociology and economics. Still yet, dem de use simulations no be only for school work but also for engineering, medicine, how to design institutions, and how to manage business.
Games too na one kind simulation system. For any game, you fit talk say di physics, di society, di rules and all inside dat game world, na like say dem de simulate dem.
Pass dat one, we de still do one kind simulation wen we de plan our life, trips, or how we go spend our pocket money.
Dem don de do dis simulations for different ways: by creating and running programs, by making and calculating equations on paper, by thinking for your head, by arranging ideas with text and arrows on a whiteboard, or by drawing graphs for Excel.
If you develop a simulation program for a particular problem, e go allow for more complex simulations pass analytical equations. But, e go need programming skills, effort, and time.
Also, di simulation model need to clear, and dat one go need skills, effort, and time for thinking.
Plus, you only fit do simulations for ways wey programs fit express, and te te now, na only wetin you fit express by calculation fit be simulated.
Generative AI go change dis situation well well.
Generative AI no only make am easy to develop simulation system programs, but also, by putting generative AI inside simulation systems, elements wey you no fit express with mathematical formulas fit also be simulated. Dis one make am possible to do vague qualitative simulation elements and simulations wey involve human-like intelligent agents.
Asides dat, dem fit express such simulation models no only with mathematical formulas but also with natural language and generative AI fit interpret am.
Dis one go make am easy to arrange di different different simulations wey we don do for different situations into systems.
Dis one go make us fit get more accurate, efficient, and effective simulation results, wey go reduce di chance of oversight and wrong assumptions.
Furthermore, wen we de discuss or tink about hard problems, we go fit use simulation systems for discussion and thinking, instead of just de rely on individual thinking simulations.
Dis one de make thinking more correct and make discussions more constructive. Dis na because instead of de point out each other's sense or mistakes for thinking, discussions fit focus on clear points like di main model of di simulation, any omissions or missing elements, how dem de estimate highly uncertain parts, and which results dem de emphasize.
As simulation systems de become easy to create, di way we de tink go shift from straight thinking, wey de focus on intuition, assumptions, and other people's bad intentions or mistakes, to simulation thinking.
E be like say you de search internet for your phone during a conversation to check news sources, Wikipedia, or original sources. You no longer need to de argue forever based only on each other's memories.
During a discussion, generative AI go arrange di simulation model, simulation rules, and wetin dem agree on before from wetin dem discuss.
Di people wey de discuss just need to add or correct information and assumptions to dat model and rules, and then confirm di simulation results. Just like wen dem find a reliable news source, those simulation results fit serve as common ground for deeper discussion.
Dis one go free listeners from di time of de wonder who right or who trustworthy. Dem no go also lose focus of di main point again as dem de try to understand di hard hard words and ideas wey de show for discussions.
Dem go only need to tink about very simple things: how to evaluate uncertainty and which values to give priority to.