<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Jeffrey Ng on Medium]]></title>
        <description><![CDATA[Stories by Jeffrey Ng on Medium]]></description>
        <link>https://medium.com/@jeffrey_nsk?source=rss-8368e53f3f79------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*_avuivMES-Zpygk35SyvpA.jpeg</url>
            <title>Stories by Jeffrey Ng on Medium</title>
            <link>https://medium.com/@jeffrey_nsk?source=rss-8368e53f3f79------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 21:01:25 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@jeffrey_nsk/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Are we speeding towards AI consciousness?]]></title>
            <link>https://medium.com/@jeffrey_nsk/are-we-speeding-towards-ai-consciousness-8fbe1f2dc6bc?source=rss-8368e53f3f79------2</link>
            <guid isPermaLink="false">https://medium.com/p/8fbe1f2dc6bc</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[algorithms]]></category>
            <category><![CDATA[consciousness]]></category>
            <dc:creator><![CDATA[Jeffrey Ng]]></dc:creator>
            <pubDate>Mon, 22 Oct 2018 15:19:39 GMT</pubDate>
            <atom:updated>2018-10-22T15:20:16.334Z</atom:updated>
            <content:encoded><![CDATA[<p><em>By Jeffrey Ng, Chief Scientist, Founders Factory (</em><a href="http://www.foundersfactory.com/"><em>www.foundersfactory.com</em></a><em>)</em></p><p>(This article has previously appeared in Computing, <a href="https://www.computing.co.uk/ctg/opinion/3063664/are-we-speeding-towards-ai-consciousness">link</a>)</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/960/1*PKm9LN2V_8hC5YiDcuIV5g.jpeg" /></figure><p>The uniqueness of humans is our consciousness, our ability to think and reason. It emerged from evolutionary pressures to build internal models of the environment and ourselves. It’s long been the ambition of computer scientists to build a machine that can replicate such human neural models. In recent months, major strides have been taken to make this a reality but we’re not at the human-level yet.</p><p>Today’s machine learning capabilities are super-human (i.e. better than human) in visual recognition tasks. They are also creeping close to the level of others in the animal kingdom, such as mice or cats, for navigation. However, to bridge the gap to human-level thinking, more needs to be done to overcome the technical hurdles in current state-of-the-art Deep Neural Networks (e.g. better neural building blocks, more flexible architectures and faster training algorithms).</p><p>In this piece, I explain recent developments paving the way to AI consciousness and why it is important to our world’s future.</p><p><strong>The consciousness debate</strong></p><p>There are two philosophies of machine learning; 1. do we keep building “narrow” AI enslaved in mundane ‘paper pushing’ jobs through massive datasets containing simple labels, e.g. image and object tags, OR 2. do we push the limits of computing to a point where it can replicate human rationalisation? I feel like the latter point is where the true value of AI can be realised for businesses and consumers, mainly through the economics of building AI systems.</p><p>Today’s AI can be compared to a curious toddler regarding its learning ability. It requires a lot of human resources to create, not the least in the amount of data labelling and experimenting with new architectures. Even superhuman AI lacks the toddler’s basic ability to self-correct, to build a minimal but sufficient model of the world and its own processes to master increasingly-complex skills.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JPqvWc5AKA2OaG78-w4MqA.jpeg" /></figure><p><strong>The trajectory to ‘thinking’ machines</strong></p><p>The hurdle we face is that our current generation of machines essentially learn to “cheat”, that is to take the fastest route from data input to data output. Our current big-data approaches pushed our AI to superhuman levels by memorising by rote every input-output pair. Compare this to the human brain which instinctively works through a sequence of connective steps to come to the desired outcome. Even with huge datasets, our deep-neural-network machines can’t yet match this human-like approach.</p><p>At this moment, no computer has shown true human-level artificial intelligence. However, the evolution from early artificial intelligence in the late 1950s to today’s more complex deep learning technology has been fast and led to the current boom in AI. The likes of DeepMind, IBM Watson, OpenAI as well as exciting startups such as Iris.AI, Bloomsbury.AI, Wayve.AI, Evolution.AI are changing the industry so fast and making the architectures slicker that the future may not be too far away.</p><p>Recent exciting projects have shown significant developments in mechanistic reasoning, such as:</p><ul><li>In July, <a href="https://www.theverge.com/2018/7/4/17533898/deepmind-ai-agent-video-game-quake-iii-capture-the-flag">DeepMind revealed its AI</a> bot playing a stripped down version of ‘Capture the Flag’ where the bot not only learned how to grab the flag but developed strategies of how to protect itself. This ability to come up with increasingly complex strategies is reminiscent of mammalian intelligence.</li><li><a href="https://openreview.net/forum?id=S1Euwz-Rb">At Stanford University, Professor Drew Hudson and Dr Christopher Manning </a>found that compositional attention networks can ‘facilitate explicit and expressive reasoning’. This is a significant step in neural networks that require less training data and hand-crafting of neural architectures for reasoning tasks, while also strengthening the machine’s ability to support explainable and structured learning. This machine’s ability to choose its own reasoning steps, the number of connective steps and the information that it stores in-between steps is a big step forward. What was found is that compositional attention networks reduce the error percentage on a Q&amp;A (Question &amp; Answer) task from 2.4% to 1.1%, making this a compelling story for making logic reasoning on a broader scale.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/940/1*W_v2Qc3gWFiasgvhVTcLgw.jpeg" /></figure><ul><li>At June’s Founder’s Forum discussion on AI, Michael Graziano’s work was highlighted as a good working theory for consciousness. <a href="https://www.frontiersin.org/articles/10.3389/fpsyg.2015.00500/full">Attention schema theory</a> proposes that the attribution of awareness to other humans is a mechanism to model what other humans’ brains are focusing on. This is a necessary step in AI’s ability to interact and perform with other intelligent agents in the loop, including us.</li></ul><p>We have come a long way in developing deep learning capabilities. But as technologists, our work is never done. As we continue to investigate and experiment with new approaches, the cost of giving machines new skills for businesses of any size will come down. What then follows is a greater use of smart AI to improve relationships between businesses, employees, consumers and the world around them.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/148/1*DHoVhcjG6HXRZA_OqXuhDA.png" /><figcaption><a href="http://www.foundersfactory.com/">http://www.foundersfactory.com</a></figcaption></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8fbe1f2dc6bc" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Where do advantages lie in university AI spin outs?]]></title>
            <link>https://medium.com/founders-factory/where-do-advantages-lie-in-university-ai-spin-outs-a352982e99cb?source=rss-8368e53f3f79------2</link>
            <guid isPermaLink="false">https://medium.com/p/a352982e99cb</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[accelerator]]></category>
            <category><![CDATA[academia]]></category>
            <category><![CDATA[venture-capital]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Jeffrey Ng]]></dc:creator>
            <pubDate>Tue, 13 Jun 2017 10:03:24 GMT</pubDate>
            <atom:updated>2018-01-11T12:02:07.846Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>For the launch of Founders Factory’s partnership with OxAI, the University of Oxford’s AI Society, I shared my experience with university students about my journey from AI research to the launch of my startup, Cortexica.</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Z5wxNBpIwDyd8Q4VniWhEw.jpeg" /><figcaption>With Rodrigo Mendoza Smith, OxAI President and Henry Lane Fox, Founders Factory CEO at the University of Oxford’s Mathematical Institute for the launch of our partnership</figcaption></figure><p><strong>Henry Lane Fox, CEO Founders Factory: Can you tell a bit about your research at Imperial [College London]?</strong><br><strong>Dr Jeffrey Ng:</strong> The journey started with a Research Council UK basic technology grant of £4M to increase the competitiveness of British industry. Our particular project was to create a next generation vision system by reverse engineering the human/mammalian visual system. For context, this was early 2002 in full throes of an AI winter and the deep learning breakthroughs in computer vision were 10 years away. We believed it was a good idea to copy this amazing piece of computational machinery that exists inside our skulls, shaped by a million years of evolution to detect food, shelter, predators, etc. So, we reverse engineered the human visual cortex into software. <br> <br><strong>HLF: How did you/your investor know there was a foundation to build a company on?</strong><br><strong>JN:</strong> For one we had a real-time demonstrator using a GPU (Graphics Processing Unit that powers Xboxes and PS3) that could detect a coke can under all rotations, in light or dark conditions etc. The venture arm of Imperial organised a beauty parade, a.k.a. a workshop, to show off our demonstrator and promise of a technology to experienced entrepreneurs in the internet space. Two of these entrepreneurs decided to get involved, one as a chairman and the other as a non-executive director, which was an early signal for the investor to move forward.<br> <br><strong> HLF: Did you have a specific idea about the go to market strategy and how many pivots did you do?</strong><br><strong>JN: </strong>We carried out many market analysis sessions with senior corporate and small enterprise executives than surfaced pain-points in the industry. Following on from that, we secured some seed money from our investor to build a scaled-up version of our technology, and proceeded to put aerials on the roof of Imperial College and bring the feeds down into the data centre. Within nine months, the lure of the smart phone was too tempting to resist and we flipped the model from monitoring live video streams for logos to matching a single image against a database of millions. We pivoted through many categories of images from wine labels, paintings, cars, fashion, etc.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8iw22j9Zroird7jeD8Oeng.jpeg" /><figcaption>“I would advise not straying too far from the basics: the size of the market opportunity and the unfair advantage provided by your tech”</figcaption></figure><p><strong>HLF: Would you discourage university spinouts that are great technology looking for problems?</strong><br><strong>JN: </strong>Quite the opposite. Several years of research at world-renowned British universities bring a level of credibility and depth to the technology in conversations with investors and clients. They also bring a level of self-confidence going into benchmarking and tech comparison exercises. At the same time, it’s tempting to see nails everywhere when you have a big hammer. I would advise not straying too far from the basics: the size of the market opportunity and the unfair advantage provided by your tech.<br> <br><strong>HLF: Would you go for big corporates, e.g. silicon-valley internet players, with their long sales cycles and how long did it take you? How much due diligence and change did it involve and how much resources did it consume internally?</strong> <br><strong>JN:</strong> Definitely, it was amazing to go through the technology sourcing and due diligence process of a big tier-1 player. B2B business cycles are traditionally long but this one involved a live real-time shoot-out with five other image recognition providers and several months of backend infrastructure design and re-architecting. Not to mention, multiple flights from London to Silicon Valley, many late night calls crossing eight time zones. And the prize at the end of the rainbow was a big contract.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*c-Na2On92TjLwTQldR5LFQ.jpeg" /><figcaption>“We carried out many market analysis sessions with senior corporate and small enterprise executives than surfaced pain-points in the industry”</figcaption></figure><p><strong>HLF: What are the lessons that you learnt along the way?</strong><br><strong>JN:</strong> It’s good to keep an eye on potential exits but I would say not worrying about it and instead focussing on building great products and having a good time with people you want to work with.<br> <br><strong>HLF: Would you recommend the people in the room to do it?</strong><br><strong>JN:</strong> Definitely, there is no faster way to learn a lot of real-world skills such as fund-raising, running a business, tech development and managing people than to “do” a startup. This rollercoaster has very tall highs and deep lows but I would argue it is a lot more satisfying than years of academic research on one problem.<br> <br><em>If you are an early stage startup and want to know more about the Founders Factory accelerator programme, visit: </em><a href="https://foundersfactory.com/accelerator/">https://foundersfactory.com/accelerator/</a></p><p>@foundersfactory</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a352982e99cb" width="1" height="1" alt=""><hr><p><a href="https://medium.com/founders-factory/where-do-advantages-lie-in-university-ai-spin-outs-a352982e99cb">Where do advantages lie in university AI spin outs?</a> was originally published in <a href="https://medium.com/founders-factory">Founders Factory</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[De-bunking AI: What does it mean and what is the opportunity?]]></title>
            <link>https://medium.com/founders-factory/de-bunking-ai-what-does-it-mean-and-what-is-the-opportunity-99d9fd13de42?source=rss-8368e53f3f79------2</link>
            <guid isPermaLink="false">https://medium.com/p/99d9fd13de42</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[accelerator]]></category>
            <category><![CDATA[entrepreneurship]]></category>
            <category><![CDATA[machine-learning]]></category>
            <dc:creator><![CDATA[Jeffrey Ng]]></dc:creator>
            <pubDate>Fri, 07 Apr 2017 15:18:32 GMT</pubDate>
            <atom:updated>2018-01-11T12:04:48.998Z</atom:updated>
            <content:encoded><![CDATA[<p><em>“Really AI is just about getting computers to operate as they do in films.”</em></p><p>If you’re looking at the pitch-decks of the hottest startups today, you’ll see AI in abundance. It is presented as the magic ingredient that can solve every problem. At Founders Factory we like to take it back to basics.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*4FAA_Og44pQL2pvxVvCdjA.jpeg" /><figcaption>On stage with Henry Lane Fox during Founders Factory’s Technology Through a New Lens event</figcaption></figure><p>For us, we like to imagine AI as software that attempts to replicate the human brain’s ability to learn from data and then do tasks. The other way to think about it is, instead of getting a bunch of coders to build a complex algorithm for you to do something clever, you just give it examples of inputs and outputs and you ask the machine to learn to do that task for you at scale.</p><p>I joined Henry Lane Fox, CEO and Co-Founder at Founders Factory, at our recent ‘Technology Through A New Lens’ event to delve deeper into the topic and unpick where the opportunities lie for founders and investors alike.</p><p>Watch our chat in full here:</p><p><a href="https://www.youtube.com/watch?v=nijZVC4hnZ4">https://www.youtube.com/watch?v=nijZVC4hnZ4</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=99d9fd13de42" width="1" height="1" alt=""><hr><p><a href="https://medium.com/founders-factory/de-bunking-ai-what-does-it-mean-and-what-is-the-opportunity-99d9fd13de42">De-bunking AI: What does it mean and what is the opportunity?</a> was originally published in <a href="https://medium.com/founders-factory">Founders Factory</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>