1. Home
  2. Detail

@str8lines: #question from @str8lines @_Barrr_None_

Str8lines
Str8lines
Open In TikTok:
Region: US
Sunday 12 June 2022 22:13:14 GMT
23268
417
7
4

Music

Download

No Watermark .mp4 (6.1MB) No Watermark(HD) .mp4 (3.87MB) Watermark .mp4 (6.22MB) Music .mp3

Comments

nickbarker2560
aces and eights custom piant :
those who cabt do the thing are the first to point out the obvious lol😏 no one is 100% everytime. but you're damn close
2022-06-13 01:03:13
2
laurie_elliott
Laurie Elliott :
What are you using to do those lines?
2022-06-13 22:18:33
0
turtlemundell305
Turtle Mundell305 :
I just want to know how you come to realize you had this talent and how long did you practice this?
2022-06-25 09:36:35
0
To see more videos from user @str8lines, please go to the Tikwm homepage.

Other Videos

Meet Rogal Dorn. Primarch of 7th Legion The Imperial Fists. #40k #warhammer40k #spacemarines #aistorytelling #cats
Meet Rogal Dorn. Primarch of 7th Legion The Imperial Fists. #40k #warhammer40k #spacemarines #aistorytelling #cats
دخيلك #karbla #مشايه_الأربعين #trend #foryou #الشعب_الصيني_ماله_حل
دخيلك #karbla #مشايه_الأربعين #trend #foryou #الشعب_الصيني_ماله_حل
Gol polémico de Huracán contra Boca!!! #futbol #bocajuniors
Gol polémico de Huracán contra Boca!!! #futbol #bocajuniors
What makes an AI system truly African? It’s not enough to give it an African-sounding name or translate it into local languages. To call it homegrown, we need to own and control the key pillars that make up any AI system. 1. Architecture This refers to how the AI model is built, the structure of the neural networks, the design choices, and the purpose it serves. If we’re using pre-existing architectures from the West (like GPT or BERT) without adapting or creating models that are tailored for African languages and use-cases, then it’s not truly African. A homegrown AI should be designed with the complexities of African languages, dialects, and cultural nuances in mind, not as an afterthought. 2. Data The fuel of AI is data. If the data used to train a model comes from non-African sources, it reflects non-African realities, priorities, and biases. For an AI to be truly African, it must be trained on African data collected ethically, in diverse languages, and from across the continent’s social, economic, and geographic landscapes. This ensures the model speaks to Africa and for Africa. 3. Policy and Regulation Even if we build the model and train it on local data, if the system is governed by external terms of service, foreign cloud providers, or outside legal frameworks, we still don’t own it. African countries need their own AI policies and data governance regulations, frameworks that protect our people, data, and innovation from exploitation or manipulation. 4. Ownership This is about control. Who funds the development? Who holds the intellectual property? Who can make decisions about how the system evolves? If African institutions, researchers, and governments are not at the centre, in control of the infrastructure, the codebase, and the deployment, then it’s not ours. Ownership also means that benefits (economic, social, scientific) return to the continent, not to corporations abroad. Until these four pillars are in place, we’re not building African AI, we’re just localising someone else’s system.
What makes an AI system truly African? It’s not enough to give it an African-sounding name or translate it into local languages. To call it homegrown, we need to own and control the key pillars that make up any AI system. 1. Architecture This refers to how the AI model is built, the structure of the neural networks, the design choices, and the purpose it serves. If we’re using pre-existing architectures from the West (like GPT or BERT) without adapting or creating models that are tailored for African languages and use-cases, then it’s not truly African. A homegrown AI should be designed with the complexities of African languages, dialects, and cultural nuances in mind, not as an afterthought. 2. Data The fuel of AI is data. If the data used to train a model comes from non-African sources, it reflects non-African realities, priorities, and biases. For an AI to be truly African, it must be trained on African data collected ethically, in diverse languages, and from across the continent’s social, economic, and geographic landscapes. This ensures the model speaks to Africa and for Africa. 3. Policy and Regulation Even if we build the model and train it on local data, if the system is governed by external terms of service, foreign cloud providers, or outside legal frameworks, we still don’t own it. African countries need their own AI policies and data governance regulations, frameworks that protect our people, data, and innovation from exploitation or manipulation. 4. Ownership This is about control. Who funds the development? Who holds the intellectual property? Who can make decisions about how the system evolves? If African institutions, researchers, and governments are not at the centre, in control of the infrastructure, the codebase, and the deployment, then it’s not ours. Ownership also means that benefits (economic, social, scientific) return to the continent, not to corporations abroad. Until these four pillars are in place, we’re not building African AI, we’re just localising someone else’s system.




About

Legal