The Kaleidoscope

There were some good questions raised from the audience too

My bestie @just_tem (check out his channel, by the way) asked an interesting question about miniaturization.

With the rapid growth of AI technology, the energy demand has skyrocketed. AI models, especially those used for large-scale machine learning and deep learning, need a ton of computational power, which eats up a lot of electricity. Some industries even rely on massive power plants to run these systems. In an eco-conscious world, isn’t that a step backward? And are there ways to miniaturize this and make it more sustainable

The answer from Nemo was interesting :

Many tech companies, like Google and Microsoft, are powering their data centers using renewable energy sources such as wind, solar, and hydropower to mitigate the environmental impact of AI's energy consumption.

Other questions were also raised most of it were specifically on the Data collection part

over all it was nice and gives such a good insight about AI


@the_kaleidoscope
@the_kaleidoscope

The main discussion points were:

- What an AI is and What AI isn't:
To put their answer in an statement :

AI is like using math to give machines some of the amazing abilities humans are blessed with. But not everything that runs on its own is actually AI. Real AI learns from experience, getting better as it goes, though the way machines think and learn is different from how we do.

- The Difference between AGI and AI

AI is typically narrow and domain-specific, excelling in specialized tasks within a particular field
AGI, or General AI, is conceptual and designed to operate across a wide range of domains. But, AGI remains theoretical and it is not practical yet.

- What are the foundation's required for an AI to become applicable :

1. Data : Well Collected, labeled, cleaned and Visualized data is necessary because AI is nothing without DATA
2. Work Force : People with the skills and the eagerness to learn about that field of study. hence awarness must be created regarding that
3. High Computing Powers: the concept of AI was thought about since in the late 40's and early 50's but couldn't become because of this one and specifically essential thing that is computing powers. at that time the computing powers of computers were slow and even the idea of GPU started in the 80's hence High computing powers is as equally essential as a DAT

- AI is becoming Centralized and becoming Open source and available will this generate an opportunity or will it have a bad side?:

Generally: There are both good and bad sides to think about. The downside is that we don’t always know what kind of data the AI is being trained on—it could be biased or flawed. On the other side, making AI open-source gives enthusiasts a chance to experiment and work on it, possibly leading to breakthroughs that rival even the big, closed-off AI projects. It’s a balance that needs to be considered carefully.

- AI in the Context of Ethiopia and the current status :

We’re definitely making progress in AI, but we're still in the early stages—not quite at the "toddler" level yet, more like infants. It’s crucial to raise awareness and get more people involved in the field. To move forward, we need to focus on three key areas, working together with the government, startups, and initiatives. For example, when it comes to data, we can’t just sit back—we need to be active in gathering and managing it properly. As for the workforce, we should start educating the younger generation early, even from elementary school, to prepare them for the future of AI.

For all of y'all who registered the event is about to start

📍Bole medhanyalem, Lingo Tower 12th floor

Just had a great day at the 3rd Pan African Conference on Artificial Intelligence held at ADWA Museum. It was organized and hosted by the Ethiopian Artificial Intelligence Institute (EAII) with the motto "Empowering Africa Through AI!". Inspired by the potential of AI, there was a discussion of the establishment of an African AI institution. Aiming to leverage AI to tackle challenges such as improving breast cancer treatment  (READ THIS ARTICLE FOR FURTHER INFO).

The conference started with an opening speech from the General Director of the Ethiopian Artificial Intelligence Institute and quickly followed up by the honorable Deputy Prime Minister of Ethiopia, Temesgen Tiruneh, made a welcoming speech on how the world is moving into a new era of digitalization and Artificial Intelligence, and how our country should be following in this new era.

Right after Dagmawi Theodros /Sr. Director at Salesforce/ presented an issue regarding "The Future of Sustainable Digital Infrastructure". The presentation raised some key points on how Empowerment through AI in Africa is nothing without the necessary Digital Infrastructures like Data centers and THIS(The Images on the Blog) were shared showing that Africa is becoming the dark continent again but this time because of digital civilization. This was illustrated by a perfect  expression "The carriage without the horse is useless" the horse being the digital infrastructures.

Jean Maillard from Meta introduced Massively Multilingual Language Technologies (MMLT). These tools can process and understand multiple languages simultaneously, facilitating global communication through features like translation, cross-lingual understanding, and text generation. MMLT models, like GPT and Meta's M2M-100, are trained on massive datasets to handle hundreds of languages. They have applications in multilingual chatbots, content creation, and AI-based translation services. Meta is collaborating with UNESCO on the UNESCO LANGUAGE TRANSLATOR project to develop MMLT tools.
A panel discussion followed, covering topics like the PanAfriCON AI Initiative, AI Ethics and Security, Digital Infrastructure, and the expectations from Engineers and Entrepreneurs.

In the afternoon, research presentations were held on Natural Language Processing, AI in Cybersecurity, and AI Applications. I attended the " Natural Language Processing: research presentation and had the chance to attend the presentations on "Donut OCR - Large Language Model where it is based on GPT(3.5)" and "Semantically Corrected Amharic Speech Recognition"

Overall In my opinion it was such a good conference and a good initiative that must be done as soon as they can especially the digital infrastructure issue is critical and must be worked up on.

and

The program Host was Solomon Kassa aka TechTalk on EBS such a nice guy met him asked him for a picture. It felt great not gonna lie

#AI #ArtificialIntelligence #EAII #Africa #Ethiopia #AddisAbaba #PanAfriConAI2024 #AIEvent #TechEvent


@the_kaleidoscope