You sabi GitHub, de web service wey open-source software developers don dey use as platform to work togeda?
For recent years, pipo don dey use am to work togeda pass just open-source software, e don reach corporate software development and even oda things wey no relate to software.
Me too, I dey use GitHub to manage my own programs and di drafts of di articles wey I dey write for dis blog.
For dis article, I go look into how GitHub go fit dey use for more things apart from software, and how e go become a shared space for knowledge wey everybody fit access.
Wiki Site Generation by DeepWiki
Plenty software development tools wey dey use generative AI, dem design dem to just help human programmers. For dis tools, humans dey write de program, and AI dey give support.
But, one new kind of software development tool dey come out where humans just dey give instructions, and generative AI dey take charge of creating de program.
One of such pioneering tools wey make pipo talk about am na Devin. Some pipo say say if you bring Devin in, e be like say you don add anoda programmer to your development team. Even though dem still dey talk say human engineers need to give detailed support for am to work well, dem go surely collect such data and use am to make improvements.
De time where one normal software development team go get one human and AI programmers like Devin as team members dey come very fast.
Cognition, de pipo wey develop Devin, don also release one service wey dem call DeepWiki.
DeepWiki na service wey dey automatically generate a wiki site for each software development project on GitHub. Dis mean say an AI like Devin go read and analyze all programs and related documents of a project, and then e go create all de documentation and design specifications.
Cognition reportedly generated wiki sites for over 50,000 of de top major public software development projects on GitHub, wey anybody fit access freely.
Since dis ones na public projects, no problem dey for doin' so. Even though wiki sites fit dey generated automatically, e must don involve plenty generative AIs wey run at full capacity for a long time, and dat one go cost plenty money.
As Cognition carry dis costs, plenty public projects benefit by gettin' documentation and design specifications for free.
If data show say dis wiki sites dey useful for public projects and dem get big impact on making quality and productivity better, software development companies go likely adopt DeepWiki for their own projects.
Cognition must don invest for generating wiki sites for plenty public projects, believing say dis one go happen. Dis one show say Cognition get confidence for DeepWiki. And if dem adopt DeepWiki, Devin go automatically follow, wey go greatly speed up how AI programmers dey popular.
GitHub as a Document Sharing Platform
GitHub don become one popular and standard web service for sharing, joint editing, and keeping programs for open-source software development.
For recent years, its strong management and security features for big companies don make advanced software development companies dey use am normally.
Because of dis, pipo often dey see GitHub as a web service mainly for keeping and sharing programs. But for real, e allow pipo to share, jointly edit, and keep different documents and materials, wey no get anything to do with programs.
Na why many pipo dey use GitHub to manage documents wey dem want to edit with many pipo. Dis documents fit relate to software or no relate at all.
Also, blogs and websites na also documents wey get some kind of program inside, or dem dey structured by programs to fit publish.
So, e no strange for individuals and companies to keep blog and website content, along with programs for presentation and automatic site generation, all together as one single GitHub project.
E also possible to make such blog and website content public GitHub projects to allow joint editing.
Recently, apart from using generative AI for software development, e dey commoner to put generative AI functionalities directly inside software.
For such cases, detailed instructions for de generative AI, wey dem dey call prompts, dey embedded inside de program.
Dem fit also consider dis prompts as a type of document.
Intellectual Factory
Even though I be software engineer, I still dey write articles for my blog.
As I want make plenty pipo read dem, e dey very hard to increase de number of readers.
Of course, I fit consider writing articles to get attention or direct contact with influential pipo for advice, putting in effort and smartness.
But, if I consider my character and de stress and effort involved, I no too like aggressive promotion. On top of dat, spending time on such activities go take time away from de main parts of my work: creating programs, thinking, and writing documents.
So, recently, I decide to try one "multimedia" or "omnichannel" strategy to make my blog articles reach more pipo by putting dem out in different content formats.
Specifically, dis one involve translating Japanese articles into English and posting dem on an English blog site, and creating presentation videos to explain articles and publishing dem on YouTube.
Furthermore, apart from publishing on general blog services, I also dey consider creating my own blog site with an index of my past articles by category and linking related articles.
If I dey create all dis manually each time a new article come out, e no go make sense. So, all tasks except writing de initial Japanese article are automated using generative AI. I call dis one an Intellectual Factory.
I need to develop programs to make dis system happen.
Currently, I don already create programs wey fit fully automate translation, presentation video generation, and YouTube uploads.
Now, I dey for de process of creating basic programs for categorizing and linking existing blog articles.
Once dat one complete, and I create a program to generate my custom blog site and automatically deploy am to a web server, de initial idea of my Intellectual Factory go fully happen.
Intellectual Factory for Everybody
De drafts of my blog articles, wey dey serve as raw materials for dis Intellectual Factory, dem dey also manage am as GitHub projects. For now, dem no publicly disclose am as private projects, but I dey consider making dem public projects for de future, along with de Intellectual Factory's programs.
Furthermore, di categorization of blog articles, de linking of articles, and de video explanations of blog articles wey I dey currently develop all get de same main idea as DeepWiki.
Using generative AI, original creative works dey use am as raw materials to produce different different content. On top of dat, information and knowledge inside dis content fit connect to create wetin dem fit call a knowledge base.
De only difference be whether de raw material na a program or a blog article. And for DeepWiki and my Intellectual Factory, wey generative AI dey power, dis difference no too matter.
In oda words, if dem interpret de term "Intellectual Factory" in a general, broader sense, no be just my specific programs, DeepWiki na also a type of Intellectual Factory.
Moreover, wetin an Intellectual Factory dey produce no dey limited to translated articles in oda languages, presentation videos, or self-made blog and wiki sites.
E go likely fit convert content into all possible media and format, like short videos, tweets, manga and anime, podcasts, and e-books.
Furthermore, de content inside dis media and formats fit also dey diversified to fit different audiences, including wider multilingualization, versions for experts or beginners, and versions for adults or children.
Finally, even generating customized content on demand go be possible.
GitHub as an Intellectual Mine
De raw materials for an Intellectual Factory fit, for principle, dey stored anywhere.
But, if we look at how GitHub don become de standard for sharing, joint editing, and keeping programs for open-source projects, and say different pipo—no be just me—dey use GitHub as a place to store documents, e dey clear say GitHub fit become de main source of raw materials for Intellectual Factories.
In oda words, GitHub go become an Intellectual Mine wey humanity share, wey go dey supply raw materials to Intellectual Factories.
De phrase "shared by humanity" here dey rhyme with de idea say open-source projects na shared software asset for humanity.
De open-source philosophy wey don support GitHub go also fit well with de idea of open documents.
On top of dat, one culture of managing copyright information and licenses for each document, similar to programs, fit come out. Content wey dem automatically generate from source documents fit easily get de same license or follow de rules wey de license talk.
From de side of creating an Intellectual Factory, putting all de raw material documents for GitHub na de best.
Dis one get two benefits: de good side of fast development, as e only need to connect GitHub to de Intellectual Factory, and de ability to clearly show de functions and performance of one's own Intellectual Factory to publicly available documents, just like DeepWiki.
For de future, as different Intellectual Factories dey developed and connected to GitHub, and more pipo and companies dey manage documents on GitHub for Intellectual Factories to process, GitHub's position as an Intellectual Mine go surely become very strong.
Humanity's Shared Public Knowledge Base
With GitHub for de center, wey dey serve as an Intellectual Mine, and Intellectual Factories wey dey produce plenty different kinds of content and knowledge bases, dis whole system go create a public knowledge base wey humanity go share.
On top of dat, dis one go be a dynamic, real-time knowledge base wey go automatically grow as de number of documents wey dem publish on GitHub dey increase.
Even though dis complex, enormous knowledge base, wey get plenty knowledge, go dey beneficial to humans, e go likely be hard for us to fully bring out its potential value.
However, AI go fit fully utilize dis publicly shared knowledge base of humanity.
Veins of Public Knowledge
When such a system come to pass, different different public information go naturally gather for GitHub.
Dis one no go just be for drafts of personal blogs or company websites.
Academic insights and data, like papers wey dem never review, research ideas, experiment results, and survey findings, go also gather there.
Dis one go attract pipo wey no just want to contribute knowledge, ideas, and data for de good of all humanity, but also pipo wey want to quickly spread discoveries to get recognition.
Even academics and researchers fit see value for their work to be checked for how true, new, and impactful e be by AI, wey dem go express through different content forms, and make am recognized by "going viral," instead of waiting for de long, time-wasting peer-review process for papers.
Alternatively, if their work catch de eye of oda researchers or companies like dis, leading to joint research or funding, real benefits dey there.
On top of dat, AI own knowledge go recirculate too.
Even though generative AI dey get plenty knowledge through pre-training, e no dey actively learn by searching for unexpected connections or similar structures among dat plenty knowledge.
De same tin apply to new insights wey dey come out from connecting different pieces of knowledge.
On de oda hand, when dem dey discuss such similarities and connections with a pre-trained generative AI, e fit accurately evaluate their value.
So, by putting different pieces of knowledge into generative AI, comparing dem anyhow or thoroughly, e possible to discover unexpected similarities and valuable connections.
Of course, because of de plenty combinations, covering everytin no go be practical. However, by properly making dis process easy and automatic, e go be possible to automatically find useful knowledge from existing knowledge.
By achieving such automatic knowledge discovery and storing de discovered knowledge on GitHub, dis cycle fit seemingly continue forever.
So, inside dis Intellectual Mine, plenty undiscovered veins dey, and e go be possible to dig dem out.
Conclusion
As one shared knowledge base for humanity, like GitHub, dey form like dis and become de standard, e go likely dey used for pre-training generative AI and for ways to find knowledge like RAG.
For such a situation, GitHub itself go work like a big brain. Generative AIs go then share dis brain, spreading and increasing knowledge.
De knowledge wey go dey recorded there no go just be records of facts, new data, or classifications. E go also include knowledge wey dey act like a catalyst, wey dey help discover oda knowledge and new combinations.
I dey call such knowledge with a catalytic effect an Intellectual Crystal, or a crystal of knowledge. Dis one include, for example, new ways of thinking.
When new ways of thinking are discovered or developed, and Intellectual Crystals are added, their catalytic effect go make new combinations and structuring of knowledge possible, wey bin no possible before, and dis go lead to more new knowledge.
Sometimes, dis ones fit contain anoda Intellectual Crystal, wey go then further increase de knowledge.
Dis kind of knowledge be like mathematical inquiry, engineering development, or invention, no be scientific discovery. So, na knowledge wey dey grow purely through thinking, no be through new things wey dem observe like scientific knowledge.
And GitHub, as an Intellectual Mine, along with plenty generative AIs wey dey use am, go make such knowledge grow faster.
Dis rapidly discovered knowledge, wey dey far pass de speed of human discovery, go dey provided in a way wey Intellectual Factories go make am easy to understand.
Like dis, knowledge wey dem fit explore purely through thinking go quickly dey discovered.