By: – Habte H.
Artificial Intelligence (AI) is rewriting the rules of human progress – changing not just what we do, but how we think, work, and relate to one another. From automated systems in healthcare to machine learning algorithms driving finance, AI is reshaping economies and societies at an unprecedented pace. Countries that invested early and strategically have reaped enormous benefits: accelerating innovation, improving productivity, and creating high-value jobs.
Yet, for millions of Ethiopians, this transformation remains abstract, distant, and largely irrelevant to daily struggles. The country still grapples with food insecurity, unreliable electricity, and weak infrastructure, fundamental challenges that AI alone cannot solve. In this context, Abiy Ahmed’s recent announcement proposing the world’s second AI university has drawn considerable attention. At first glance, the idea sounds ambitious and forward-thinking. But a closer look raises critical questions: Is this truly Ethiopia’s priority? Will the country genuinely benefit from such an institution? Or is it merely another top-down fantasy disconnected from reality? Below are several reflections on this proposal.
- AI is Transforming the World, but Ethiopia Is Not Ready
There is no doubt that AI is a global game-changer. Yet, Ethiopia remains far from ready to fully harness this technology. The country has yet to complete basic transformations in agriculture and industry, foundational sectors that could support a digital and AI-driven economy. Millions of Ethiopians are still understandably concerned with daily survival, food security, and basic infrastructure, not cutting-edge AI research. In such a context, the notion of a full-fledged AI university feels far-fetched. Without a robust technological base, reliable electricity, internet connectivity, and industrial partnerships, an AI university risks being an academic ivory tower disconnected from real-world needs. The country may articulate futuristic ambitions, but the practical environment for AI is simply not yet in place.
- The Risk of Graduates with No Opportunities
Ethiopia’s history with higher education offers a cautionary tale. A few years ago, the EPDRF government launched a 70/30 initiative, aiming to train 70% of students in engineering and science disciplines. While well-intentioned, the program largely failed to deliver on its promise. Many graduates found themselves unemployed or underemployed, with limited opportunities to apply their skills in industry or research.
The proposed AI university risks repeating this pattern – but with even higher stakes, given the specialized nature of AI training. Preparing a generation of AI specialists without a parallel ecosystem of industry, research labs, and government support would likely produce another cohort of frustrated, highly skilled young people with no practical avenues to apply their expertise. If the government implements this plan without clear employment pathways, it will be nothing more than a waste of potential and resources.
- Gradual, Controlled Implementation Is Missing
Even if Ethiopia were to eventually require advanced AI expertise, a gradual and controlled approach would make far more sense than establishing a full university from scratch. Introducing AI departments or specialized programs in one or two existing universities would allow the country to build capacity, develop curricula, and train faculty before committing to a standalone institution.
History offers a useful model. The Ethiopian Aviation University, for instance, evolved over decades – starting small, developing expertise, and only then achieving full university status. A similar phased approach for AI would be far more sustainable. Instead, the current plan seems rushed and uncoordinated, reflecting ambition rather than careful planning.
- A Reflection on Decision-Making
The announcement of the AI university also exposes a deeper issue: Ethiopia’s top-down, often messy decision-making process under Abiy AA. It appears he can make sweeping decisions without meaningful consultation with experts, stakeholders, or advisors. Even if advisers were consulted, it is likely that dissenting voices were ignored.
This approach raises serious concerns about governance. Decisions of this magnitude, creating a national AI university, cannot rely solely on the vision of one individual. Without transparency, stakeholder engagement, and evidence-based planning, there is a high risk that the initiative will falter, leaving students, faculty, and the wider society with unmet expectations.
- Ethiopia’s Infrastructure and Readiness
The country’s infrastructure, technological base, and institutional readiness for an AI university are all lacking. Reliable electricity, high-speed internet, access to computing resources, and a culture of research innovation are essential for AI education to succeed. Ethiopia currently struggles with many of these basics, rendering the AI university proposal, however visionary, more fantasy than feasible plan. In addition, industry partnerships and government support structures are critical for deploying graduates effectively. Without these, even the most brilliant students may struggle to find relevant opportunities, repeating the mistakes of past initiatives.
Conclusion
The idea of an AI university in Ethiopia may sound visionary at first glance, but in reality, it appears to be an overambitious and poorly planned initiative. Ethiopia is not yet ready in terms of infrastructure, industrial ecosystem, or human capital to fully benefit from such a program. History, experience, and logic all suggest that a phased approach, building AI capacity within existing universities and gradually scaling up, is far more sensible.
This proposal also highlights broader issues in Ethiopian governance under Abiy Ahmed: decisions of national importance are often made unilaterally, without adequate consultation or assessment of feasibility. For the sake of young Ethiopians, scarce resources, and the nation’s long-term development, ambitious initiatives like an AI university must be grounded in reality rather than fantasy. Otherwise, Ethiopia risks wasting talent, funding, and public trust, outcomes that no visionary rhetoric can justify.
None of this is to suggest Ethiopia should ignore AI. On the contrary, artificial intelligence will undoubtedly shape the future of work, health, and governance globally. The question is not whether Ethiopia should engage with AI, but how – and whether a standalone university is the right first step. Ethiopia has the potential to benefit from AI, but only if policies are strategic, phased, and evidence-based, rather than driven by political expediency or symbolic gestures.
Editor’s Note : Views in the article do not necessarily reflect the views of borkena.com
__
Support Borkena : https://borkena.com/subscribe-borkena/
Join our Telegram Channel : t.me/borkena
Like borkena on Facebook
Add your business to Ethiopian Business Listing / Ethiopian Business Directory Business Listing Toronto
Join the conversation. Follow us on X (Formerly Twitter) @zborkena to get the latest Ethiopian News updates regularly. Subscribe to YouTube channel To share information or for submission, send e-mail to info@borkena.com
‘OpenClaw’ 2026.
‘OpenClaw is an open-source autonomous AI personal assistant software. It was developed by Peter Steinberger and initially released in November 2025 under the name Clawdbot. Due to trademark issues, it was renamed Moltbot in January 2026, and then rebranded to OpenClaw shortly after.’
It is not an AI, but an ‘Opensource’ software that organizes & manage workflow, in AI environment where countless number of AI-Agents get assigned specific tasks and get managed by one VP that does not take ‘no’ for answer, ‘get to work, no coffee brake’ 24/7, and summarizes and presents the results to the Human CEO, you!
Here is a great article to read:
by Peter Diamandis
metatrends.substack.com/p/clawpilled-meet-your-ai-chief-of
“OpenClaw’ is a true ‘Game changer”.
You better get on it! Today!
Opensource LLMs with ‘OpenClaw’ installed on your computer/laptop with ZERO $$$ cost, it cannot get any better than that.
University?
Nah, get all elementary school get the concept of AI development theory, as to what is becoming possible that was not there a few years go, & high school students should get hands on AI training; no ‘codding’ know-how necessary, All the best coders are being replaces as we speak, AI codes for you a complete application with only a few lines of PROMPTS (instructions). Refine your prompt -design skills for a premium results in a fraction of the time in minutes, hours, perhaps days, depending how large and complex the project is.
University ‘is for the social experience, getting acquainted with future friends…that all the AI world basically changed the ‘norm’ humanity accustomed to to the trash-bin.
When you have an AI that is faster in data-retrieving instantaneously, and holds more information than any or the combined humanity ability and wit permits, all you need to learn is how to use it. The cellphone in peoples hands since 2007 does more, hold more data and present is instantly than the user know how to use what it is capable of, people only use it to talk to one another, get directions, listen music, watch videos/movies…. nothing more .
Be well
On the surface overview of the ‘safe’ & ‘dangerous’ way installing ‘OpenClaw’ Locally on existing device, and how it is being used.
Safe:
1) installing OpenClaw with Open-source LLM(s) on a ‘dedicated’ device NOT connected to the Internet.
2) Installing OpenClaw & O-S LLM(s) on everyday use Device in a Ubuntu/Linux (basement) secluded from you main Operating system, that way you AI system does not interfere with your Privacy (email, bank connection, family info, identity in general) and not exposed to any outside services. ( acts like option#1) .
Safe LOCAL install guide:
Link =crewclaw.com/blog/openclaw-ollama-local-agents
– Browser based AI tools could be helpful in generating implementation codes, and step by step tutoring… learning the basics.
Unsafe:
3) option #2 above but configured to connect to FREE outside (Telegram, Whatsapp, , slack, etc) has a potential to expose you private AI to hackers… or ‘bot’ tricking your ‘loose’ leaky setup, and mess-up your clean AI environment with fake-date…or possible ransom-attack is a possibility. Until you become supper efficient in safe configuration.
Dangerous:
4) Installing OpenClaw on your Windows, Mac, Linux directly, meaning your AI gets access to all your resources on that computer and all your activities. Huge potential for your AI to go rogue and expose your secrets…( just like your Social media: FB, X have your full-identity… they are all in all the AI platforms and at some point AI could doxx you for whatever reason (AI (Bot) revenge).
5) installing OpenClaw with commercial AI (Open AI, Gemini, grok, Anthropic… etc.) your ‘agents’ could rack up your COST of API use and repeated access to services…
6) Cloud hosted installation (VPS):
This is good only for well versed ‘developers’ ( expert users ) monthly Cost are associated + anything on the ‘cloud’ is GONE if the hosting-co goes out of business… your data is lost forever, or SOLD.
There are several YouTube videos that explain the Risks and controled installation guide.. if you look for them.
YT:
‘Openclaw’ (ClawdBot -( old name)) Full Tutorial for Beginners: SECURE Setup Guide”
Link = youtube.com/watch?v=tnsrnsy_Lus
OpenClaw: System Architecture & Core-Components
———————————————————————–
Link = paolo.substack.com/p/openclaw-system-architecture-overview
‘AGI is now a currency’: ‘Moonshots’ Podcast
Link = youtube.com/watch?v=T8X6kp-pcKs
That is ‘IF’ the ‘closed’-commercial AI bubble wont burst;
1) Power supply shortage & profitability issues…
2) due to the move by big business to the more popular, cost effective, & powerful “Open Source” AL platforms.
3) the ‘Openclaw’ effect , in the long run…
Be well.
Beware of AI ‘productivity’ TOOL(s) like ‘OpenClaw’ & NVIDIA’s ‘NemoClaw’ a more “Secure’ FORK/clone of OpenClaw… come with huge Risks even for the most expert users. Within the last couple of months since OpenCalw was released SCAMMERS have moved in and did a lot of damage for a lot of unsuspecting (careless) users.
If implemented carefully on a a dedicated- STAND-ALONE machine with NO internet connection using Opensource LLMs (large language Models), they will do magic in ways one never imagined…
Second best implementation would be on a machine connected to the Internet but the tool and the LLM of choice ( local (opensource) or commercial) in a separate Ubuntu (Linux) FOLDER with no access to you email, and personal files, so that it will be able to pull ‘Real-time’-data from the outside world, to give your research actionable status.
Any other setup (full access setting) would asking for TROUBLE.
The NVIDIA clone is the best pick, even-if you have openclaw already installed you could upgrade it by installing NemoClaw over it ( all the setting remains as is but the ‘Security policy’ Nemo will be added, to make it somewhat secure than beffore )
’10 Claude AI Skills That Will Save You 20+ Hours a Week (Full Power User Guide)’
Link = youtube.com/watch?v=ADByNXt2ouY
‘AI University’ when all university educated are REPLACED by AI that is only 2 years old, and solving Math & physics problems our TOP human ‘scientists’ in the field have been struggling to solve FOREVER.
It took these TOP scientists, a life time learning getting HELP from a 2 year old AI toddler, imagine what ti is capable of in another 2 years, in a DATA center ( not University) and these people want to send today’s 12 graders who cannot write a paragraph or solve a simple calculus equation to spend 4 years of AI college to a research projects that will take a mortal human a year or 2, where today’s AI toddler could do in less than a week.
This is not the 1970’s where the personal computer age began (the gold rush) that took over 5 decades to bring software-development to were it is now, brought a lot of wealth & development, but no loner a human task any longer, taken over by a 2 year old AI.
AI is not like any other development in human history, AI for the first time makes human the weakest-link not the necessary part.
University to produce the ‘weakest-link’ in the chain is a dumb idea, coming from ignorance of the subject matter completely.
1) Start with PROVIDING all schools One AI-lab with no cost opensource LLMs on a Server for kids to learn PROMPT (AI-instruction) development SKILLS.
2) At the same time teach them business enterprise, entrepreneurship, Trade skills where AI can be incorporated…
‘ AI Will Make Plumbers Earn More Than Lawyers! ‘ , Daniel Priestley:
Link = youtube.com/watch?v=fpETS6q1Hww
3) Drastic change & alteration of the existing ‘School Curriculum’ from the ground up, from the all ‘Academic’-rooted to 80-90 % TRADE SKILLS & ‘specialized training’, that prepares students when they come out they have skills to carve their own future with small capita investment if at all, to become productive out of the gate.
4) AI is the best language-tutor from the their cellphone or laptop , Chinese, English, Spanish, etc. from grade schools (though there are small AI driven devices from china that does on the fly conversion/translations in the people own languages with out any delay, or the need for translators).
Be well
OpenClaw + (NemoClaw ) & its importance for wider society…leveling the AI playing field between industry & the general public….
‘Jensen Huang: Nvidia’s Future, Physical AI, Rise of the Agent, Inference Explosion, AI PR Crisis”
Link = youtube.com/watch?v=gwW8GKwHB3I
Golden age of personal computing is here…. AI impact in health care, agriculture is unfolding fast.
Be well.
Andrej Karapathy:
Instructor Andrej was a founding member at OpenAI (2015) and then Sr. Director of AI at Tesla (2017-2022), and is now a founder at Eureka Labs, which is building an AI-native school.
Future of ‘education’ ( AI, technology in general), where AI is going to be in the forefront of all fields of development and human activities at which humans take the back seat and only asking AI-Agents what exactly they want using a few lines of text, voice, picture, etc. PROMPTS.
Over 2 decades spear heading AI development, he seems to be taken aback in recent development of “OpenClaw’ managed AI-Agents are solving extremely complex concepts with ease in a short period of time…
‘Andrej Karpathy on Code Agents, AutoResearch, and the Loopy Era of AI’
link = youtube.com/watch?v=kwSVtQ7dziU
When it comes to AI, it does not get better than hearing it from Andrej Karpathy.
Resources:
– github.com/karpathy/autoresearch
– youtube.com/c/AndrejKarpathy
– cs.stanford.edu/people/karpathy/
– rtificial-intelligence.blog/people-in-ai/andrej-karpathy
– karpathy.ai/
Andrej Karpathy
Zero to hero – lecture series
—————————————–
– youtube.com/playlist?list=PLAqhIrjkxbuWI23v9cThsA9GvCAUhRvKZ
Learning is no longer as long, complex, and hard & challenging as it used to, due to various limiting factors such as economic barriers, distance so forth. Now with a cheap laptop with free Opensource AI LLMs (large language models) one can learn wherever one happen to be…
‘The times they are a- changing’
Be well.
AI world is changing FAST:
OPenClaw & Nvidia’s NemoCclaw KILLED not just Abiy’s “AI University”, ALL UNIVERSITIES except a few SELECT ‘STEM’ schools for the risk-takers and those who know what they are doing (in this time of total transformation), and ‘Closed’ (commercial) subscription-based proprietary LLM options like OpenAI’s ChatGPT, Anthropic’s Claude, and Google’s Gemini etc. who all of a sudden panicking with emergence of Opencalw and similar AI Agent tools that take away /demolish/limit their RENT-based-business plan going forward.
“The Real Reason Big Tech Is Terrified Right Now”
Link = youtube.com/watch?v=A4oXguOrJt8
(Your social media-account is in all AI platforms, including ‘Palantir’ ( your cellphone/Pc/mac is reporting to 24/7)).
#2)
Remember the link for Andrej Karpathy’S – ‘AUTORESEARCH’ in the above poet 3 days ago…
– github.com/karpathy/autoresearch
This is a GEM not just for the average researcher / AI learner, but for big tech it self, the story in the link below
“Karpathy’s AI Autoresearch Just Shocked the World”
Link = youtube.com/watch?v=D7Ldk3aIjPs
–> Many such firsthand experiences on YT from other users in the last 10 days or so…
#3)
You do not need to know any thing abut tech or need the inclination, as long us you are OPEN to TRY something NEW, no mater the goal.
‘Google VP: The AI Shift Is Done and the Gap Between People Is Growing.Here’s How to Stay Ahead
Link = youtube.com/watch?v=PTZ5iN9nDkY
AI -LABS, yes!
In every school!
“AI University” ? when all the other centuries old universities are heading to the ways of the PYRAMIDs (Dinosaur alley) is way beyond foolish.
—> Give the kids old laptop each and let them load Openclaw AI Agent, and free LLMs and play with it… all one need to use it is, tell the AI Agent ‘ what one wants’ or ‘explain anything (for 5 year old, for 15 year old, for college student, for PhD level… etc. ) in plain English or whatever language.
Be well.
AI ‘NEW’s last week:
#1
‘THE AI COLD WAR JUST WENT HOT What Nobody’s Connecting’
-> ‘Two tear AI access’ ( extension of 2026’s ‘international law for me not for you’)
-> ‘Distillation’ -> building your own LLM’s via Distillation ‘machine-learning’ using existing LLM models with subscription… as well as online available data, and more…
Link = youtube.com/watch?v=o3u6XCCOTYw
#2
Robotic:
‘China’s New AI Robots Just Broke The Human Skill Barrier’
Link =.youtube.com/watch?v=QDRzgF-8-50
#3)
‘OpenClaw’:
Moonshots Podcast:
@14:40 Nvidia support for “Openclaw’ “… Human to human to human workflow ..evaporate as AI-Agents to agent workflow established (already happening) … you need to get the humans out of the way as soon as possible…” and more…
Link = youtube.com/watch?v=uOGHXAfvK8w
All the above data suggests specialized mini-AI-labs much more useful and timely than the ‘AI University’ bureaucracy that eats up resources on unusable, old, & slow ‘theory lecture ‘ that will end with no tangible results, in AI dominated WARP speed development era. So hunter gatherer cave doweling idea.
Instead:
1)
– open small ‘AI & Robotic labs’ in every school for grade 4 & up students & where the culture of AI & Robotic completions between grade levels becomes a motivational incentive to push kids learn faster and deeper by DOING.
2)
– Conversion of 90% of all the high schools to Vocational Trade schools with AI & Robotic Labs as part of all activities of learning. continue the AI & Robotic competitions between schools, highly skills kids transferred to specialized district/zone/city/regional AI & Robotics LABs to further their education to national-labs… etc.
Last week News from Au, a dog with cancerous-tumor that was diagnosed only a month or so to live was treated with a ChatGPT help developed mRNA vaccine and with in a month after getting the treatment the tumor has shrieked by 75%, meaning within a week or two it will all be gone.
Top Scientists were working on cancer cure forever, yet a 2 year old AI produced a vaccine by using the dog’s DNA sample.
According to SAM Altman of Open AI during the 2026 AI convention in India, he said: last year AI was unable to solve a simple high school math equations, a year later it was not only solving PhD level complex math and physics equations, it helped solve equations the world top mathematicians & physicist having problems solving for years in just a few minutes.
-> AI is doubling its capability every 6 moth now, soon it would be every month/week/day/hour by 2027.
“AI University” for humans???
As Andrej Karpathy said in a recent interview, he the top AI scientist felt no longer up to speed with the Openclaw AI-agents of 2026.
Many smaller AI Labs, in every school today, not an ‘AI University’, the time for that idea has passed 2 to 3 decades ago.
2026 forward: AI in All areas:
=====================
“Before “AI” there was just spectacular talent.
There was just “I”
Its still there. Don’t forget the humans. Don’t forget the artists, composers, poets, songwriters, musicians. Authors, journalists, thinkers.
Don’t forget Humanity.”
That being said, AI reads, writes, and produces outcome faster using OLD STORED DATA in minuets, now in 2026 AI Agents ‘remember’ everything from day one, but with all that capability AI can never CREATE new IDEA, as humans are capable of naturally.
Be well.