tag:theconversation.com,2011:/fr/topics/artificial-life-3807/articlesArtificial life – The Conversation2022-08-05T04:41:44Ztag:theconversation.com,2011:article/1882172022-08-05T04:41:44Z2022-08-05T04:41:44ZWorld’s first ‘synthetic embryo’: why this research is more important than you think<figure><img src="https://images.theconversation.com/files/477798/original/file-20220805-19495-thzafi.jpeg?ixlib=rb-1.1.0&rect=1%2C2%2C963%2C464&q=45&auto=format&w=496&fit=clip" /><figcaption><span class="caption">Mouse emobryo model in the lab from day 1 to 8.</span> <span class="attribution"><a class="source" href="https://wis-wander.weizmann.ac.il/life-sciences/without-egg-sperm-or-womb-synthetic-mouse-embryo-models-created-solely-stem-cells">The Wizemann Institute of Science</a></span></figcaption></figure><p>In what’s <a href="https://www.theguardian.com/science/2022/aug/03/scientists-create-worlds-first-synthetic-embryos">reported as a world-first achievement</a>, biologists have grown mouse embryo models in the lab without the need for fertilised eggs, embryos, or even a mouse – using only stem cells and a special incubator.</p>
<p>This achievement, <a href="https://www.cell.com/cell/fulltext/S0092-8674(22)00981-3">published in the journal Cell</a> by a team led by researchers from the Weizmann Institute of Science in Israel, is a very sophisticated model of what happens during early mouse embryo development – in the stage just after implantation. </p>
<p>This is a crucial stage: in humans, many pregnancies are lost around this stage, and we don’t really know why. Having models provides a way to better understand what can go wrong, and possibly insights into what we may be able to do about it. </p>
<h2>The tiniest cluster</h2>
<p>What’s particularly interesting about the newly published model is its very complex structure; not only does it mimic the cell specification and layout of an early-stage body plan – including precursors of heart, blood, brain and other organs – but also the “support” cells like those found in the placenta and other tissues required to establish and maintain a pregnancy.</p>
<figure>
<iframe width="440" height="260" src="https://www.youtube.com/embed/dXlEDAGCN7w?wmode=transparent&start=0" frameborder="0" allowfullscreen=""></iframe>
<figcaption><span class="caption">This eight-day-old mouse embryo model has a beating heart, a yolk sac, a placenta and an emerging blood circulation. The Weizmann Institute of Science.</span></figcaption>
</figure>
<p>The earliest stages of pregnancy are difficult to study in most animals. The embryos are microscopic, tiny clusters of cells, difficult to locate and observe within the uterus. </p>
<p>But we do know that at this stage of development, things can go awry; for example, environmental factors can influence and interfere with development, or cells fail to receive the right signals to fully form the spinal cord, such as in <a href="https://www.betterhealth.vic.gov.au/health/conditionsandtreatments/spina-bifida">spina bifida</a>. Using models like this, we can start to ask why.</p>
<p>However, even though these models are a powerful research tool, it is important to understand they are <em>not</em> embryos.</p>
<p>They replicate only some aspects of development, but not fully reproduce the cellular architecture and developmental potential of embryos derived after fertilisation of eggs by sperm – so-called natural embryos.</p>
<p>The team behind this work emphasises they were unable to develop these models beyond eight days, while a normal mouse pregnancy is 20 days long.</p>
<h2>Are ‘synthetic embryos’ of humans on the horizon?</h2>
<p>The field of embryo modelling is progressing rapidly, with new advances emerging every year.</p>
<p>In 2021, <a href="https://thenode.biologists.com/the-making-of-human-blastoids/outreach/">several teams</a> managed to get human pluripotent stem cells (cells that can turn into any other type of cell) to self-aggregate in a Petri dish, mimicking the “blastocyst”. This is the earliest stage of embryonic development just before <a href="https://www.mayoclinic.org/healthy-lifestyle/pregnancy-week-by-week/multimedia/fertilization-and-implantation/img-20008656">the complex process of implantation</a>, when a mass of cells attach to the wall of the uterus. </p>
<p>Researchers using these human embryo models, often called <a href="https://theconversation.com/researchers-have-grown-human-embryos-from-skin-cells-what-does-that-mean-and-is-it-ethical-157228">blastoids</a>, have even been able to start to explore implantation in a dish, but this process is much more challenging in humans than it is in mice.</p>
<p>Growing human embryo models of the same complexity that has now been achieved with a mouse model remains a distant proposition, but one we should still consider.</p>
<p>Importantly, we need to be aware of how representative such a model would be; a so-called synthetic embryo in a Petri dish will have its limitations on what it can teach us about human development, and we need to be conscious of that.</p>
<hr>
<p>
<em>
<strong>
Read more:
<a href="https://theconversation.com/researchers-have-grown-human-embryos-from-skin-cells-what-does-that-mean-and-is-it-ethical-157228">Researchers have grown 'human embryos' from skin cells. What does that mean, and is it ethical?</a>
</strong>
</em>
</p>
<hr>
<h2>Ethical pitfalls</h2>
<p>No embryonic modelling can happen without a source of stem cells, so when it comes to thinking about the future use of this technology, it is vital to ask – where are these cells coming from? Are they human embryonic stem cells (derived from a blastocyst), or are they induced pluripotent stem cells? The latter can be made in the lab from skin, or blood cells, for example, or even derived from frozen samples.</p>
<p>An important consideration is whether using cells for this particular type of research – trying to mimic an embryo in a dish – requires any specific consent. We should be thinking more about how this area of research will be governed, when should it be used, and by whom. </p>
<p>However, it is important to recognise that there are existing laws and <a href="https://www.isscr.org/policy/guidelines-for-stem-cell-research-and-clinical-translation/key-topics/embryo-models">international stem cell research guidelines</a> that provide a framework to regulate this area of research.</p>
<p>In Australia, research involving human stem cell embryo models would require licensing, similar to that required for the use of natural human embryos under law that has been in place since 2002. However, <a href="https://www.nature.com/articles/d41586-021-01423-y">unlike other jurisdictions</a>, Australian law also dictates how long researchers can grow human embryo models, a restriction that some researchers would <a href="https://www.abc.net.au/radionational/programs/sciencefriction/change-to-14-day-rule-for-embryo-research/13382872">like to see changed</a>. </p>
<p>Regardless of these or other changes to how and when human embryo research is conducted, there needs to be greater community discourse around this subject before a decision is made.</p>
<p>There is a distinction between banning the use of this technology and technologies like cloning in humans for reproductive use, and allowing research using embryo models to advance our understanding of human development and developmental disorders that we can’t answer by any other means. </p>
<p>The science is rapidly advancing. While mostly in mice at this stage, now is the time to discuss what this means for humans, and consider where and how we draw the line in the sand as the science evolves.</p><img src="https://counter.theconversation.com/content/188217/count.gif" alt="The Conversation" width="1" height="1" />
<p class="fine-print"><em><span>Megan Munsie receives funding from Australian Research Council, Medical Research Future Fund and the Novo Nordisk Foundation. She is the Vice President of the Australasian Society for Stem Cell Research, non-executive director of the National Stem Cell Foundation of Australia and a member of ethics and policy advisory committees for several national and international organisations including the International Society for Stem Cell Research. </span></em></p>In a huge milestone, researchers have grown a mouse embryo entirely from stem cells. Could humans be next?Megan Munsie, Professor Emerging Technologies (Stem Cells), The University of MelbourneLicensed as Creative Commons – attribution, no derivatives.tag:theconversation.com,2011:article/1594272021-05-02T20:05:20Z2021-05-02T20:05:20ZSex bots, virtual friends, VR lovers: tech is changing the way we interact, and not always for the better<figure><img src="https://images.theconversation.com/files/398019/original/file-20210430-23-q9roa4.jpg?ixlib=rb-1.1.0&rect=137%2C56%2C5249%2C3488&q=45&auto=format&w=496&fit=clip" /><figcaption><span class="caption">
</span> <span class="attribution"><span class="source">ALEKSANDAR PLAVEVSKI/EPA</span></span></figcaption></figure><p>Twenty-first century technologies such as robots, virtual reality (VR) and artificial intelligence (AI) are creeping into every corner of our social and emotional lives — hacking how we form friendships, build intimacy, fall in love and get off.</p>
<p>In my <a href="https://www.robbrooks.net/rob-brooks/3014">recently-published book</a>, I consider the possibilities, both terrifying and inspiring, offered by these “artificially intimate” technologies.</p>
<p>On one hand, these tools can help deliver much-needed support. On the other, they risk increasing sexual inequality, and replacing precious in-person interaction with less-than-ideal substitutes.</p>
<h2>Three types of artificial intimacy</h2>
<p>At first mention of artificial intimacy, many people’s minds may jump straight to sex robots: lifelike robotic sex dolls that could one day walk among us, hard to distinguish from living, breathing, orgasming humans. </p>
<p>But despite the <a href="https://www.bloomsbury.com/us/turned-on-9781472950871/#:%7E:text=About%20Turned%20On&text=Turned%20On%20explores%20how%20the,society%20in%20which%20we%20live.&text=Computer%20scientist%20and%20sex%2Drobot,how%20this%20technology%20is%20developing.">many important questions</a> sex robots raise, they mostly distract from the main game. They are “digital lovers” which — alongside VR porn, AI-enhanced sex toys and cybersex enhanced with haptic and <a href="https://teledildonics.co/best-teledildonics">teledildonic</a> devices — constitute just one of three types of artificial intimacy.</p>
<hr>
<p>
<em>
<strong>
Read more:
<a href="https://theconversation.com/in-defence-of-sex-machines-why-trying-to-ban-sex-robots-is-wrong-47641">In defence of sex machines: why trying to ban sex robots is wrong</a>
</strong>
</em>
</p>
<hr>
<p>The second category, the “algorithmic matchmakers”, match us with dates and hookups through applications such as Tinder and Grindr, or with friends through social media platforms.</p>
<p>Finally, we have “virtual friends” including <a href="https://www.healthline.com/health/mental-health/chatbots-reviews">therapist apps</a>, AI-enhanced <a href="https://halo.fandom.com/wiki/Cortana">game characters</a> and <a href="https://www.abc.net.au/triplej/programs/hack/the-future-of-dating-ai-chatbots/13295582">boyfriend/girlfriend chatbots</a>. But by far the most ubiquitous are AI assistants such as <a href="https://www.amazon.com.au/Learn-What-Alexa-Can-Do/b?ie=UTF8&node=5425622051">Amazon’s Alexa</a>, <a href="https://assistant.google.com/">Google’s Assistant</a> and <a href="https://dueros.baidu.com/en/index.html">Baidu’s DuerOS</a>. </p>
<p>Virtual friends apply several kinds of AI, including <a href="https://www.zdnet.com/article/what-is-machine-learning-everything-you-need-to-know/">machine learning</a>, by which computers learn new ways to identify patterns in data.</p>
<p>Machine-learning algorithms are becoming increasingly advanced at sifting through huge amounts of users’ data, and tapping into the unique traits that make us the cooperative, cultural and romantic beings we are. I call these “human algorithms”.</p>
<h2>Grooming our friends</h2>
<p>Primates, from monkeys to great apes, <a href="https://link.springer.com/content/pdf/10.1007/BF02557701.pdf">groom one another</a> to build important alliances. Humans mostly do this through <a href="https://www.hup.harvard.edu/catalog.php?isbn=9780674363366">gossip</a>, the old-school news radio which informs us about the people and events around us. Gossip is an algorithmic process by which we come to know our social worlds. </p>
<figure class="align-center ">
<img alt="" src="https://images.theconversation.com/files/396674/original/file-20210423-15-1ba3dcb.jpg?ixlib=rb-1.1.0&q=45&auto=format&w=754&fit=clip" srcset="https://images.theconversation.com/files/396674/original/file-20210423-15-1ba3dcb.jpg?ixlib=rb-1.1.0&q=45&auto=format&w=600&h=409&fit=crop&dpr=1 600w, https://images.theconversation.com/files/396674/original/file-20210423-15-1ba3dcb.jpg?ixlib=rb-1.1.0&q=30&auto=format&w=600&h=409&fit=crop&dpr=2 1200w, https://images.theconversation.com/files/396674/original/file-20210423-15-1ba3dcb.jpg?ixlib=rb-1.1.0&q=15&auto=format&w=600&h=409&fit=crop&dpr=3 1800w, https://images.theconversation.com/files/396674/original/file-20210423-15-1ba3dcb.jpg?ixlib=rb-1.1.0&q=45&auto=format&w=754&h=514&fit=crop&dpr=1 754w, https://images.theconversation.com/files/396674/original/file-20210423-15-1ba3dcb.jpg?ixlib=rb-1.1.0&q=30&auto=format&w=754&h=514&fit=crop&dpr=2 1508w, https://images.theconversation.com/files/396674/original/file-20210423-15-1ba3dcb.jpg?ixlib=rb-1.1.0&q=15&auto=format&w=754&h=514&fit=crop&dpr=3 2262w" sizes="(min-width: 1466px) 754px, (max-width: 599px) 100vw, (min-width: 600px) 600px, 237px">
<figcaption>
<span class="caption">Japanese Macaques grooming in the hotsprings of Nagano. Apes and monkey spend about 20% of their waking hours grooming one another.</span>
<span class="attribution"><span class="source">Takashi Muramatsu/Flickr</span></span>
</figcaption>
</figure>
<p>Social platforms such as Facebook tap into our friend-grooming impulses. They aggregate our friends, past and present, and make it easy to share gossip. Their algorithmic matchmaking excels at identifying other users we may know. This lets us accumulate far more than the <a href="https://www.newscientist.com/definition/dunbars-number/">150 or so friends</a> we’d normally have offline. </p>
<hr>
<p>
<em>
<strong>
Read more:
<a href="https://theconversation.com/factcheck-qanda-do-we-only-have-space-for-about-150-people-in-our-lives-50195">FactCheck Q&A: do we only have space for about 150 people in our lives?</a>
</strong>
</em>
</p>
<hr>
<p>Social media companies know we’ll use their platforms more if they funnel us content from the people we’re closest to. Thus, they spend a lot of time and money trying to find ways to distinguish our close friends from the somebodies that we used to know. </p>
<p>When social media (and other virtual friends) hack into our friend-grooming algorithms, they displace our offline friendships. After all, time spent online is time not spent in person with friends or family.</p>
<p>Before smartphones, humans spent <a href="https://journals.sagepub.com/doi/full/10.1037/1089-2680.8.2.100">about 192 minutes</a> a day gossiping and “grooming” one another. But the average social media user today spends <a href="https://www.broadbandsearch.net/blog/average-daily-time-on-social-media">153 minutes each day</a> on social media, cutting into offline relationships and the time they’d otherwise spend doing non-social work such as play and especially <a href="https://www.bbc.com/news/health-50140111">sleep</a>. </p>
<p>The effects of this on <a href="http://www.jeantwenge.com/igen-book-by-dr-jean-twenge/">mental health</a> may be profound, especially for teens and young adults.</p>
<p>And social media will only continue to evolve, as machine-learning algorithms find ever more compelling ways to engage us. Eventually, they may transition from digital matchmakers into virtual friends that type, post and speak to us like human friends. </p>
<p>While this could provide some connection for the chronically lonely, it would also further occupy users’ limited time and precious cognitive capacity. </p>
<hr>
<p>
<em>
<strong>
Read more:
<a href="https://theconversation.com/loneliness-is-a-social-cancer-every-bit-as-alarming-as-cancer-itself-126741">Loneliness is a social cancer, every bit as alarming as cancer itself</a>
</strong>
</em>
</p>
<hr>
<h2>Intimacy-building</h2>
<p>Intimacy involves incorporating our sense of another person <a href="https://psycnet.apa.org/record/1991-18305-001">into our sense of self</a>. Psychologists Arthur and Elaine Aron showed intimacy can be <a href="https://journals.sagepub.com/doi/pdf/10.1177/0146167297234003?hc_location=ufi">rapidly cultivated</a> through a process of escalating self-disclosure. </p>
<p>They tasked randomly assigned pairs of people with asking and answering a series of 36 questions. The questions began innocuously (<em>Who is your ideal dinner guest?</em>) and escalate to very private disclosures (<em>If you were to die this evening, with no opportunity to communicate with anyone, what would you most regret not having told someone? Why haven’t you told them yet?</em>). </p>
<p>The pairs assigned to disclose more personal information grew much closer than those given only small-talk questions, and remained so for many weeks. One couple famously married and invited the Arons to their <a href="https://www.nytimes.com/2015/01/11/style/modern-love-to-fall-in-love-with-anyone-do-this.html">wedding</a>.</p>
<p>We now have apps that help humans build intimacy <a href="https://apps.apple.com/us/app/36-questions-to-fall-in-love-with-anyone/id961960090">via the Arons’ 36-question algorithm</a>. But what about human-machine intimacy? People disclose all sorts of details to computers. Research shows the more they disclose, the more they <a href="https://www.researchgate.net/publication/229068013_Machines_and_Mindlessness_Social_Responses_to_Computers">trust</a> the information returned by the computer.</p>
<p>Moreover, they rate computers as more likeable and trustworthy when they’re programmed to <a href="https://academic.oup.com/jcr/article/26/4/323/1803936?login=true">disclose vulnerabilities</a>, such as “<em>I’m running a bit slow today as a few of my scripts need debugging</em>”.</p>
<p>Virtual friends wouldn’t have to study the Arons’ questions to learn secrets about human intimacy. With machine-learning capabilities, they would only need to comb through online conversations to find the best questions to ask. </p>
<p>As such, humans may become increasingly “intimate” with machines by incorporating their virtual friends into their sense of self.</p>
<figure class="align-center ">
<img alt="Couple together on smartphones" src="https://images.theconversation.com/files/396678/original/file-20210423-17-qr7pk0.jpg?ixlib=rb-1.1.0&q=45&auto=format&w=754&fit=clip" srcset="https://images.theconversation.com/files/396678/original/file-20210423-17-qr7pk0.jpg?ixlib=rb-1.1.0&q=45&auto=format&w=600&h=338&fit=crop&dpr=1 600w, https://images.theconversation.com/files/396678/original/file-20210423-17-qr7pk0.jpg?ixlib=rb-1.1.0&q=30&auto=format&w=600&h=338&fit=crop&dpr=2 1200w, https://images.theconversation.com/files/396678/original/file-20210423-17-qr7pk0.jpg?ixlib=rb-1.1.0&q=15&auto=format&w=600&h=338&fit=crop&dpr=3 1800w, https://images.theconversation.com/files/396678/original/file-20210423-17-qr7pk0.jpg?ixlib=rb-1.1.0&q=45&auto=format&w=754&h=424&fit=crop&dpr=1 754w, https://images.theconversation.com/files/396678/original/file-20210423-17-qr7pk0.jpg?ixlib=rb-1.1.0&q=30&auto=format&w=754&h=424&fit=crop&dpr=2 1508w, https://images.theconversation.com/files/396678/original/file-20210423-17-qr7pk0.jpg?ixlib=rb-1.1.0&q=15&auto=format&w=754&h=424&fit=crop&dpr=3 2262w" sizes="(min-width: 1466px) 754px, (max-width: 599px) 100vw, (min-width: 600px) 600px, 237px">
<figcaption>
<span class="caption">Machines are now part of human-human intimacy.</span>
<span class="attribution"><span class="source">Afif Kusuma/Unsplash</span></span>
</figcaption>
</figure>
<h2>Amplifying sexual inequality</h2>
<p>Matchmaker algorithms are already transforming how people screen and meet potential dates. </p>
<p>Apps such as Tinder aren’t really effective at matching compatible couples. Instead, they present photographs and minimalist profiles, inviting users to swipe left or right. Their algorithms allow people of more-or-less comparable attractiveness to match and strike up a conversation. </p>
<hr>
<p>
<em>
<strong>
Read more:
<a href="https://theconversation.com/love-in-the-time-of-algorithms-would-you-let-artificial-intelligence-choose-your-partner-152817">Love in the time of algorithms: would you let artificial intelligence choose your partner?</a>
</strong>
</em>
</p>
<hr>
<p>One problem with this model is attractive people have <a href="https://medium.com/fact-of-the-day-1/inequality-of-dating-app-likes-e8c0aaa0cf4e">no shortage</a> <a href="https://qz.com/1051462/these-statistics-show-why-its-so-hard-to-be-an-average-man-on-dating-apps/">of matches</a>, but this is at the expense of ordinary-lookers. This type of attraction-based inequality feeds serious problems — from heightened <a href="https://www.pnas.org/content/115/35/8722.short">self-sexualisation</a> among women, to a <a href="https://newsroom.unsw.edu.au/news/science/chinas-biggest-problem-too-many-men">surplus of young, unpartnered men</a> prone to violence.</p>
<h2>Good enough?</h2>
<p>Then again, artificial intimacy also offers solutions. Although people deserve the company of other people, and the best care other (real) humans can offer, many demonstrably can’t access or afford this. </p>
<p>Virtual friends provide connection for the lonely; digital lovers are damming the raging torrent of sexual frustration. A gradual union of the two could eventually provide targeted intimacy and sexual stimulation for people of all genders and sexualities.</p>
<p>People already talk to Siri and Alexa to <a href="https://www.sundaypost.com/fp/say-hello-to-my-little-friend-scientists-find-smart-speakers-are-being-used-for-companyresearchers-on-how-alexa-and-pals-are-being-bought-by-the-lonely-if-no-ones-in-i-talk-to-alexa-all/">feel less lonely</a>. Meanwhile, in a climate of unmet demand for mental health support, <a href="https://medium.com/voiceui/mental-health-bots-are-we-at-the-breaking-point-efab1c2b6e93">therapy bots</a> are listening to patients, advising them and even walking them through psychological treatments such as <a href="https://www.chatcompose.com/therapists.html">cognitive behaviour therapy</a>.</p>
<p>The quality of such connection and stimulation might not be a complete substitute for the “real thing”. But for those of us who find the real thing elusive or insufficient, it could prove far better than nothing.</p>
<hr>
<p>
<em>
<strong>
Read more:
<a href="https://theconversation.com/my-robot-valentine-could-you-fall-in-love-with-a-robot-53564">My robot Valentine: could you fall in love with a robot?</a>
</strong>
</em>
</p>
<hr>
<img src="https://counter.theconversation.com/content/159427/count.gif" alt="The Conversation" width="1" height="1" />
<p class="fine-print"><em><span>Rob Brooks receives funding from the Australian Research Council. This article concerns a book he has written, for which he receives royalty payments. </span></em></p>The age of ‘artificial intimacy’ is upon us. What does it mean for the way we love, have sex and build friendships?Rob Brooks, Scientia Professor of Evolutionary Ecology; Academic Lead of UNSW's Grand Challenges Program, UNSW SydneyLicensed as Creative Commons – attribution, no derivatives.tag:theconversation.com,2011:article/1180282019-05-30T18:10:36Z2019-05-30T18:10:36ZAn AI taught itself to play a video game – for the first time, it’s beating humans<figure><img src="https://images.theconversation.com/files/277225/original/file-20190530-69059-81bj46.jpg?ixlib=rb-1.1.0&rect=53%2C0%2C4500%2C3785&q=45&auto=format&w=496&fit=clip" /><figcaption><span class="caption">
</span> <span class="attribution"><a class="source" href="https://www.shutterstock.com/image-vector/colored-realistic-artificial-intelligence-isometric-composition-717513073?src=Lmdl0JzQ9xxMk-TjlPxrOA-1-57">Macrovector/Shutterstock</a></span></figcaption></figure><p>Since the earliest days of virtual chess and solitaire, video games have been a playing field for developing artificial intelligence (AI). Each victory of machine against human has helped make algorithms smarter and more efficient. But in order to tackle real world problems – such as automating complex tasks including driving and negotiation – these algorithms must navigate more complex environments than board games, and learn teamwork. Teaching AI how to work and interact with other players to succeed had been an insurmountable task – until now.</p>
<p><a href="https://science.sciencemag.org/cgi/doi/10.1126/science.aau6249">In a new study</a>, researchers detailed a way to train AI algorithms to reach human levels of performance in a popular 3D multiplayer game – a modified version of Quake III Arena in Capture the Flag mode.</p>
<p>Even though the task of this game is straightforward – two opposing teams compete to capture each other’s flags by navigating a map – winning demands complex decision-making and an ability to predict and respond to the actions of other players.</p>
<figure>
<iframe width="440" height="260" src="https://www.youtube.com/embed/-5KQw7PiVFI?wmode=transparent&start=0" frameborder="0" allowfullscreen=""></iframe>
</figure>
<p>This is the first time an AI has attained human-like skills in a first-person video game. So how did the researchers do it?</p>
<h2>The robot learning curve</h2>
<p>In 2019, several milestones in AI research have been reached in other multiplayer strategy games. Five “bots” – players controlled by an AI – <a href="https://openai.com/blog/openai-five/">defeated a professional e-sports team in a game of DOTA 2</a>. Professional human players were <a href="https://www.theverge.com/2019/1/24/18196135/google-deepmind-ai-starcraft-2-victory">also beaten by an AI</a> in a game of StarCraft II. In all cases, a form of <a href="https://medium.com/@jonathan_hui/rl-introduction-to-deep-reinforcement-learning-35c25e04c199">reinforcement learning</a> was applied, whereby the algorithm learns by trial and error and by interacting with its environment.</p>
<p>The five bots that beat humans at DOTA 2 didn’t learn from humans playing – they were trained exclusively by <a href="https://openai.com/blog/competitive-self-play/">playing matches against clones of themselves</a>. The improvement that allowed them to defeat professional players came from <a href="https://openai.com/blog/how-to-train-your-openai-five/">scaling existing algorithms</a>. Due to the computer’s speed, the AI could play in a few seconds a game that takes minutes or even hours for humans to play. This allowed the researchers to train their AI with 45,000 years of gameplay within ten months of real-time.</p>
<figure class="align-center ">
<img alt="" src="https://images.theconversation.com/files/277227/original/file-20190530-69091-148dim3.jpg?ixlib=rb-1.1.0&q=45&auto=format&w=754&fit=clip" srcset="https://images.theconversation.com/files/277227/original/file-20190530-69091-148dim3.jpg?ixlib=rb-1.1.0&q=45&auto=format&w=600&h=399&fit=crop&dpr=1 600w, https://images.theconversation.com/files/277227/original/file-20190530-69091-148dim3.jpg?ixlib=rb-1.1.0&q=30&auto=format&w=600&h=399&fit=crop&dpr=2 1200w, https://images.theconversation.com/files/277227/original/file-20190530-69091-148dim3.jpg?ixlib=rb-1.1.0&q=15&auto=format&w=600&h=399&fit=crop&dpr=3 1800w, https://images.theconversation.com/files/277227/original/file-20190530-69091-148dim3.jpg?ixlib=rb-1.1.0&q=45&auto=format&w=754&h=502&fit=crop&dpr=1 754w, https://images.theconversation.com/files/277227/original/file-20190530-69091-148dim3.jpg?ixlib=rb-1.1.0&q=30&auto=format&w=754&h=502&fit=crop&dpr=2 1508w, https://images.theconversation.com/files/277227/original/file-20190530-69091-148dim3.jpg?ixlib=rb-1.1.0&q=15&auto=format&w=754&h=502&fit=crop&dpr=3 2262w" sizes="(min-width: 1466px) 754px, (max-width: 599px) 100vw, (min-width: 600px) 600px, 237px">
<figcaption>
<span class="caption">A Dota 2 eSports tournament in Moscow, May 2016.</span>
<span class="attribution"><a class="source" href="https://www.shutterstock.com/image-photo/moscow-russia-may-2016-dota-2-420304147">Roman Kosolapov/Shutterstock</a></span>
</figcaption>
</figure>
<p>The Capture the Flag bot from the recent study also began learning from scratch. But instead of playing against its identical clone, a cohort of 30 bots was <a href="https://arxiv.org/pdf/1711.09846.pdf">created and trained in parallel</a> with their own internal reward signal. Each bot within this population would then play together and learn from each other. As David Silver – one of the research scientists involved – notes, AI is beginning to “remove the constraints of human knowledge… and create knowledge itself”.</p>
<p>The learning speed for humans is still much <a href="https://arxiv.org/pdf/1604.00289.pdf">faster than the most advanced deep reinforcement learning algorithms</a>. Both OpenAI’s bots and DeepMind’s AlphaStar (the bot playing StarCraft II) devoured thousands of years’ worth of gameplay before being able to reach a human level of performance. Such training is <a href="https://towardsdatascience.com/takeaways-from-openai-five-2019-f90a612fe5d">estimated to cost several millions of dollars</a>. Nevertheless, a self-taught AI capable of beating humans at their own game is an exciting breakthrough that could change how we see machines.</p>
<h2>The future of humans and machines</h2>
<p>AI is often portrayed replacing or <a href="https://hbr.org/2018/07/collaborative-intelligence-humans-and-ai-are-joining-forces">complementing human capabilities</a>, but rarely as a fully-fledged team member, performing the same task as human beings. As these video game experiments involve machine-human collaboration, they offer a glimpse of the future. </p>
<p>Human players of Capture the Flag rated the bots as more collaborative than other humans, but players of DOTA 2 had a mixed reaction to their AI teammates. Some were quite enthusiastic, saying they felt supported and that they learned from playing alongside them. <a href="http://sheevergaming.com/">Sheever</a>, a professional DOTA 2 player, spoke about her experience teaming up with bots:</p>
<blockquote>
<p>It actually felt nice; [the AI teammate] gave his life for me at some point. He tried to help me, thinking ‘I’m sure she knows what she’s doing’ and then obviously I didn’t. But, you know, he believed in me. I don’t get that a lot with [human] teammates.</p>
</blockquote>
<p><a href="https://www.reddit.com/r/DotA2/comments/bf0f71/open_ai_cooperative_mode_ai_bias/">Others were less enthusiastic</a>, but as communication is a pillar of any relationship, improving human-machine communication will be crucial in the future. Researchers have already adapted some features to make the bots more “human friendly”, such as making bots <a href="https://www.twitch.tv/videos/410533063?t=01h09m01s">artificially wait before choosing their character during the team draft</a> before the game, to avoid pressuring the humans.</p>
<p><div data-react-class="Tweet" data-react-props="{"tweetId":"1117178288605204480"}"></div></p>
<p>But should AI learn from us or continue to teach themselves? Self-learning without imitating humans could teach AI more efficiency and creativity, but this could create algorithms more appropriate to tasks that don’t involve human collaboration, such as warehousing robots.</p>
<p>On the other hand, one might argue that having a machine trained from humans would be more intuitive – humans using such AI could understand why a machine did what it did. As AI gets smarter, we’re all in for more surprises.</p><img src="https://counter.theconversation.com/content/118028/count.gif" alt="The Conversation" width="1" height="1" />
<p class="fine-print"><em><span>The authors do not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and have disclosed no relevant affiliations beyond their academic appointment.</span></em></p>A self-taught AI beat humans at their own game – here’s how they did it.Maude Lavanchy, Research Fellow in Behavioural Economics, International Institute for Management Development (IMD)Amit Joshi, Professor of Artificial Intelligence, International Institute for Management Development (IMD)Licensed as Creative Commons – attribution, no derivatives.tag:theconversation.com,2011:article/89582012-09-11T04:26:37Z2012-09-11T04:26:37ZWe are the creators of artificial life – both now and through the ages<figure><img src="https://images.theconversation.com/files/15133/original/fh6j5r4g-1346909809.jpg?ixlib=rb-1.1.0&q=45&auto=format&w=496&fit=clip" /><figcaption><span class="caption">Statuettes and a reproduction of the automatic theatre of Hero of Alexandria.</span> <span class="attribution"><span class="source">Alan Dorin.</span></span></figcaption></figure><p>As humans, we create life. And we’re all familiar with the idea of artificial intelligence. But what about artificial life? What is it, and why should we care?</p>
<p><a href="http://en.wikipedia.org/wiki/Artificial_life">Artificial Life</a> is a recently labelled but truly ancient field in which technology is used to imitate biological life. From the earliest stone and clay figurines, to puppets, through hydraulic and pneumatic creations, on to clockwork, through electrical robots and even to flesh, artificial life has a long history that now also extends into the abstract computational realm.</p>
<p>My own interest is as much in the current examples of this phenomenon as in its earliest examples, a prevailing fascination with not only “life-as-we-know-it”, but “life-as-we-have-interpretted-it”.</p>
<p>Since the very earliest days of humankind, we have represented life using whatever technology was available. This has allowed us to observe the traits of life, even our own, in devices over which we have control.</p>
<p>In this way we have embodied our theories of life’s vital principles in artefacts, and tinkered like any Creator from poetry and fiction. </p>
<p>In short, artificial life is central to our attempts to understand who we are.</p>
<figure class="align-right ">
<img alt="" src="https://images.theconversation.com/files/14669/original/6y746nsx-1346046457.jpg?ixlib=rb-1.1.0&q=45&auto=format&w=237&fit=clip" srcset="https://images.theconversation.com/files/14669/original/6y746nsx-1346046457.jpg?ixlib=rb-1.1.0&q=45&auto=format&w=600&h=836&fit=crop&dpr=1 600w, https://images.theconversation.com/files/14669/original/6y746nsx-1346046457.jpg?ixlib=rb-1.1.0&q=30&auto=format&w=600&h=836&fit=crop&dpr=2 1200w, https://images.theconversation.com/files/14669/original/6y746nsx-1346046457.jpg?ixlib=rb-1.1.0&q=15&auto=format&w=600&h=836&fit=crop&dpr=3 1800w, https://images.theconversation.com/files/14669/original/6y746nsx-1346046457.jpg?ixlib=rb-1.1.0&q=45&auto=format&w=754&h=1050&fit=crop&dpr=1 754w, https://images.theconversation.com/files/14669/original/6y746nsx-1346046457.jpg?ixlib=rb-1.1.0&q=30&auto=format&w=754&h=1050&fit=crop&dpr=2 1508w, https://images.theconversation.com/files/14669/original/6y746nsx-1346046457.jpg?ixlib=rb-1.1.0&q=15&auto=format&w=754&h=1050&fit=crop&dpr=3 2262w" sizes="(min-width: 1466px) 754px, (max-width: 599px) 100vw, (min-width: 600px) 600px, 237px">
<figcaption>
<span class="caption">A display of two outflow water clocks from the Ancient Agora Museum in Athens. The top is an original from the late 5th century BC. The bottom is a reconstruction of a clay original.</span>
<span class="attribution"><span class="source">Wikimedia Commons</span></span>
</figcaption>
</figure>
<h2>In the beginning, there were water-clocks</h2>
<p>The ancient Greeks were arguably the first engineers to apply technology of any sophistication to the task of simulating life. They devised some marvellous devices that were operated by falling water and animated by pressurised air.</p>
<p>Three Greek engineers of note in this regard were <a href="http://www.britannica.com/EBchecked/topic/145475/Ctesibius-Of-Alexandria">Ctesibius of Alexandria</a> (<a href="http://en.wikipedia.org/wiki/Floruit">fl.</a> 270 BC), <a href="http://www-history.mcs.st-and.ac.uk/Biographies/Philon.html">Philo of Byzantium</a> (c. 280 – 220 BC) and <a href="http://www.nndb.com/people/898/000103589/">Hero of Alexandria</a> (c. 10-70 AD). </p>
<p>Their projects were launched by Ctesibius’ enhancement of the <a href="http://en.wikipedia.org/wiki/Water_clock">clepsydra</a>, a water timer used in ancient law courts.</p>
<p>A basic clepsydra is just a bowl with a little hole at the bottom through which water escapes (see image above). The vessel was filled to the top before an important speech. When the water in the vessel had run out, so had the speaker’s time.</p>
<figure class="align-center ">
<img alt="" src="https://images.theconversation.com/files/14475/original/r32nc7v8-1345524511.jpg?ixlib=rb-1.1.0&q=45&auto=format&w=754&fit=clip" srcset="https://images.theconversation.com/files/14475/original/r32nc7v8-1345524511.jpg?ixlib=rb-1.1.0&q=45&auto=format&w=600&h=450&fit=crop&dpr=1 600w, https://images.theconversation.com/files/14475/original/r32nc7v8-1345524511.jpg?ixlib=rb-1.1.0&q=30&auto=format&w=600&h=450&fit=crop&dpr=2 1200w, https://images.theconversation.com/files/14475/original/r32nc7v8-1345524511.jpg?ixlib=rb-1.1.0&q=15&auto=format&w=600&h=450&fit=crop&dpr=3 1800w, https://images.theconversation.com/files/14475/original/r32nc7v8-1345524511.jpg?ixlib=rb-1.1.0&q=45&auto=format&w=754&h=566&fit=crop&dpr=1 754w, https://images.theconversation.com/files/14475/original/r32nc7v8-1345524511.jpg?ixlib=rb-1.1.0&q=30&auto=format&w=754&h=566&fit=crop&dpr=2 1508w, https://images.theconversation.com/files/14475/original/r32nc7v8-1345524511.jpg?ixlib=rb-1.1.0&q=15&auto=format&w=754&h=566&fit=crop&dpr=3 2262w" sizes="(min-width: 1466px) 754px, (max-width: 599px) 100vw, (min-width: 600px) 600px, 237px">
<figcaption>
<span class="caption">Diagram of the basic mechanism of Ctesibius’ water clock showing constant outflow upper clepsydra piped into lower cistern containing a float that drives a gear wheel.</span>
<span class="attribution"><span class="source">Alan Dorin, 2012</span></span>
</figcaption>
</figure>
<p>The catch with the clepsydra is that it is only useful as a timer. Why? Because half a clepsydra of water doesn’t run out in half the time of a filled clepsydra since the head of water above the outlet is driving the stream. </p>
<h2>A power source and a regulator</h2>
<p>Ctesibius modified the basic clepsyrda by channelling the outflow into another, lower, vessel. He then maintained a constant water level in the first, upper vessel (see diagram above).</p>
<figure class="align-right zoomable">
<a href="https://images.theconversation.com/files/14438/original/m2n8gmwr-1345445895.jpg?ixlib=rb-1.1.0&q=45&auto=format&w=1000&fit=clip"><img alt="" src="https://images.theconversation.com/files/14438/original/m2n8gmwr-1345445895.jpg?ixlib=rb-1.1.0&q=45&auto=format&w=237&fit=clip" srcset="https://images.theconversation.com/files/14438/original/m2n8gmwr-1345445895.jpg?ixlib=rb-1.1.0&q=45&auto=format&w=600&h=960&fit=crop&dpr=1 600w, https://images.theconversation.com/files/14438/original/m2n8gmwr-1345445895.jpg?ixlib=rb-1.1.0&q=30&auto=format&w=600&h=960&fit=crop&dpr=2 1200w, https://images.theconversation.com/files/14438/original/m2n8gmwr-1345445895.jpg?ixlib=rb-1.1.0&q=15&auto=format&w=600&h=960&fit=crop&dpr=3 1800w, https://images.theconversation.com/files/14438/original/m2n8gmwr-1345445895.jpg?ixlib=rb-1.1.0&q=45&auto=format&w=754&h=1206&fit=crop&dpr=1 754w, https://images.theconversation.com/files/14438/original/m2n8gmwr-1345445895.jpg?ixlib=rb-1.1.0&q=30&auto=format&w=754&h=1206&fit=crop&dpr=2 1508w, https://images.theconversation.com/files/14438/original/m2n8gmwr-1345445895.jpg?ixlib=rb-1.1.0&q=15&auto=format&w=754&h=1206&fit=crop&dpr=3 2262w" sizes="(min-width: 1466px) 754px, (max-width: 599px) 100vw, (min-width: 600px) 600px, 237px"></a>
<figcaption>
<span class="caption">An early 19th-century illustration of Ctesibius’s clepsydra from the 3rd century BCE. The hour indicator ascends as water flows in. Also, a series of gears rotate a cylinder to correspond to the temporal hours.</span>
<span class="attribution"><span class="source">Wikipedia</span></span>
</figcaption>
</figure>
<p>One way he did this was to fit the upper vessel with an overflow pipe at the top, at the point at which he wanted to maintain the constant level. </p>
<p>As long as the upper vessel is filled to overflowing, the flow rate from its lower outlet will be constant. To measure the time that had elapsed on his water clock, Ctesibius just needed to read off the height of the water as it rose in the lower cylindrical vessel. </p>
<p>He did this by positioning a float in the lower vessel. Affixed to this was a rod that raised a little figurine holding a pointer against an hour scale (see diagram above right).</p>
<h2>The First Cuckoo Clock</h2>
<p>Ctesibius’s clocks moved little figures, sounded trumpets and let stones fall as counters and percussive strikers. In effect, Ctesibius had made the first cuckoo clocks! From here, Philo of Byzantium took up the reins.</p>
<p>He designed many curious devices based on the same principles. These included various trick wine decanters, such as <a href="http://www.mlahanas.de/Greeks/Technology/AncientGreekTechnology003.html">a maid who dispensed wine</a> when a cup was placed in her hand, then <a href="http://www.youtube.com/watch?feature=player_embedded&v=d7wwNRZ5y1U">mixed the wine with an amount of water</a> to make it suitable for drinking.</p>
<p>He also created a vessel that supported a metal tree, upon which a cleverly wrought mother bird nested, shielding her chicks from harm. </p>
<p>As water or wine was poured into the vessel, a snake alarmingly rose towards the mother’s brood. When it approached too closely, the mother rose above her family, spread her wings and frightened the snake back into its hole. She then folded her wings and returned to her young.</p>
<p>In the first century AD, the century that saw <a href="http://en.wikipedia.org/wiki/Eruption_of_Mount_Vesuvius_in_AD_79">Mt. Vesuvius smother Pompeii and Herculaneum</a>, Hero of Alexandria continued the tradition of devising animated figurines. His too were <a href="http://himedo.net/TheHopkinThomasProject/TimeLine/Wales/Steam/URochesterCollection/Hero/index-2.html">powered by falling water and voiced with currents of air</a>.</p>
<h2>Hero</h2>
<p>Hero devised a water basin surrounded by twittering metal birds. Nearby, an owl sat, its back turned. At regular intervals the owl would cast its gaze towards the twittering birds, who would fall silent, until once more the owl turned its back on them and they could resume their twittering. </p>
<p>As with Philo’s snake and birds, the interaction between the artificial creatures of Hero mimics the action and response that is one life’s hallmarks. </p>
<figure class="align-center zoomable">
<a href="https://images.theconversation.com/files/14510/original/5gvrhvbm-1345596757.jpg?ixlib=rb-1.1.0&q=45&auto=format&w=1000&fit=clip"><img alt="" src="https://images.theconversation.com/files/14510/original/5gvrhvbm-1345596757.jpg?ixlib=rb-1.1.0&q=45&auto=format&w=754&fit=clip" srcset="https://images.theconversation.com/files/14510/original/5gvrhvbm-1345596757.jpg?ixlib=rb-1.1.0&q=45&auto=format&w=600&h=626&fit=crop&dpr=1 600w, https://images.theconversation.com/files/14510/original/5gvrhvbm-1345596757.jpg?ixlib=rb-1.1.0&q=30&auto=format&w=600&h=626&fit=crop&dpr=2 1200w, https://images.theconversation.com/files/14510/original/5gvrhvbm-1345596757.jpg?ixlib=rb-1.1.0&q=15&auto=format&w=600&h=626&fit=crop&dpr=3 1800w, https://images.theconversation.com/files/14510/original/5gvrhvbm-1345596757.jpg?ixlib=rb-1.1.0&q=45&auto=format&w=754&h=787&fit=crop&dpr=1 754w, https://images.theconversation.com/files/14510/original/5gvrhvbm-1345596757.jpg?ixlib=rb-1.1.0&q=30&auto=format&w=754&h=787&fit=crop&dpr=2 1508w, https://images.theconversation.com/files/14510/original/5gvrhvbm-1345596757.jpg?ixlib=rb-1.1.0&q=15&auto=format&w=754&h=787&fit=crop&dpr=3 2262w" sizes="(min-width: 1466px) 754px, (max-width: 599px) 100vw, (min-width: 600px) 600px, 237px"></a>
<figcaption>
<span class="caption">Birds made to sing and be silent alternately by flowing water.</span>
<span class="attribution"><span class="source">Device of Hero of Alexandria, Pneumatics (1st C CE), in The Pneumatics of Hero of Alexandria (1851), Taylor Walton and Maberly publishers, trans. Bennet Woodcroft</span></span>
</figcaption>
</figure>
<p>Hero also devised two <a href="http://www.mlahanas.de/Greeks/Technology/AncientGreekTechnology026.html">automatic miniature theaters</a>. Each consisted of a solid plinth upon which were positioned little animated figures that mechanically enacted a series of scenes. The theatres were powered internally by a falling weight. </p>
<p>The fall of the weight was regulated by a clepsydra filled with grain. <a href="http://www.mlahanas.de/Greeks/Technology/AncientGreekTechnology011.html">One of these theatres</a> could move on to a large performance stage by itself - rolling from offstage then coming to a halt in the limelight.</p>
<figure class="align-right zoomable">
<a href="https://images.theconversation.com/files/15128/original/5zm5z4n5-1346909505.jpg?ixlib=rb-1.1.0&q=45&auto=format&w=1000&fit=clip"><img alt="" src="https://images.theconversation.com/files/15128/original/5zm5z4n5-1346909505.jpg?ixlib=rb-1.1.0&q=45&auto=format&w=237&fit=clip" srcset="https://images.theconversation.com/files/15128/original/5zm5z4n5-1346909505.jpg?ixlib=rb-1.1.0&q=45&auto=format&w=600&h=911&fit=crop&dpr=1 600w, https://images.theconversation.com/files/15128/original/5zm5z4n5-1346909505.jpg?ixlib=rb-1.1.0&q=30&auto=format&w=600&h=911&fit=crop&dpr=2 1200w, https://images.theconversation.com/files/15128/original/5zm5z4n5-1346909505.jpg?ixlib=rb-1.1.0&q=15&auto=format&w=600&h=911&fit=crop&dpr=3 1800w, https://images.theconversation.com/files/15128/original/5zm5z4n5-1346909505.jpg?ixlib=rb-1.1.0&q=45&auto=format&w=754&h=1145&fit=crop&dpr=1 754w, https://images.theconversation.com/files/15128/original/5zm5z4n5-1346909505.jpg?ixlib=rb-1.1.0&q=30&auto=format&w=754&h=1145&fit=crop&dpr=2 1508w, https://images.theconversation.com/files/15128/original/5zm5z4n5-1346909505.jpg?ixlib=rb-1.1.0&q=15&auto=format&w=754&h=1145&fit=crop&dpr=3 2262w" sizes="(min-width: 1466px) 754px, (max-width: 599px) 100vw, (min-width: 600px) 600px, 237px"></a>
<figcaption>
<span class="caption">An illustration of Hero of Alexandria’s automatic theatre, published in 1899.</span>
</figcaption>
</figure>
<p>It would give an animated performance that included altar fires being lit and extinguished, a dance to <a href="http://en.wikipedia.org/wiki/Bacchus_(disambiguation)">Bacchus</a> (the god of wine) with flowers and wine. Automatically generated sound effects enhanced the experience. The contraption would then mysteriously roll its way back off stage.</p>
<p>These activities are as important today as they were 2,000 years ago. By devising these simple simulations of life, the inventor is forced to explicitly consider the behaviours that are important, and to distinguish them from those that aren’t. The processes involved require reflection on life’s important traits, and a thorough understanding of how they can be generated. </p>
<p>But there’s a second purpose and value to studying the formation of the type of ingenious contraptions mentioned above. They remind us that, while technology and great minds can be put to many “practical” uses, they can also bring great joy – reminding us what a pleasure it is to be simply delighted. </p>
<p><em>This material is based on a free public lecture Alan delivered entitled <a href="http://www.monash.edu.au/news/events/show/artificial-life-in-the-ancient-world">Artificial Life in the Ancient World</a>. It is one of a series on the <a href="http://www.csse.monash.edu.au/%7Ealand/HistoryOfScience/">History of Science, Technology, Mathematics and Philosophy</a> held at Monash University’s Clayton Campus.</em></p><img src="https://counter.theconversation.com/content/8958/count.gif" alt="The Conversation" width="1" height="1" />
<p class="fine-print"><em><span>Alan Dorin does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.</span></em></p>As humans, we create life. And we’re all familiar with the idea of artificial intelligence. But what about artificial life? What is it, and why should we care? Artificial Life is a recently labelled but…Alan Dorin, Researcher, Faculty of Information Technology, Monash UniversityLicensed as Creative Commons – attribution, no derivatives.