Task-Based Language Teaching And Its Impact On Oral Presentation: Fill & Download for Free

The following is more of a diary entry recounting my CVPR experiences together with the papers I found most interesting.Sorry for the bits (industrial exhibit, workshops) that are irrelevant to the question.All opinions expressed here are my own and not endorsed by present or past employers, collaborators, and, friends.—-Today was the last day of CVPR and associated workshops. This trip as always was an overwhelming experience, so much so, that at one point I literally called 911 and had to get the fire department to rescue me! But more on that later.Since uploading papers on arXiv right after submission (for many, even before submission) became a thing – the main conference has lost its freshness. I was already aware of many of the kickass ideas in problems that I am interested in. Fortunately there were still a few pleasant surprises.I’ll start by talking about the papers that I particularly liked, and end with observations on the rest of the program.Main Conference: the papersMost of the work that interested me can be filed under the following headings:1. Artificial IntelligenceOverall CVPR seems to have transformed from a computer vision conference to an AI conference. A significant number of the papers presented weren’t talking exclusively about vision anymore.In particular, the program was flooded with Visual Question Answering (VQA, vision + language) papers with a significant representation in the orals. To be honest, I was initially skeptical of these works. I thought of such attempts as chickening out of the really challenging problems of vision – as nobody expects strong results on entirely novel problems, and they are presumably easier to publish. However now, with ever increasing recognition accuracies, as we start saturating tag-level tasks, I feel that we will be holding ourselves back if we don’t start leveraging more sophisticated language constructs to describe visual concepts. On this, I was particularly moved by a talk by Ivan Laptev (at a workshop) where he pointed out the incredible diversity found in crowdsourced labels obtained for any given video snippet [0]. We communicate concepts via language, and the visual concepts we model can only be as sophisticated as the language constructs we use to describe them. My favorite capability tangentially related to VQA was demonstrated in “Counting Everyday Objects in Everyday Scenes” [1] – though I don’t really understand the approach yet.Another paper that falls in the broad AI category is titled “Image to Image Translation” [2], and uses adversarial training to learn a variety of generative tasks in an unsupervised manner. The funny thing about this paper is that it has the same idea as Apple’s paper that won the Best Paper Award at the conference – whereas this paper didn’t even get a spotlight or oral presentation! I personally liked [2] better because it discusses many more applications and results. For that matter, the other best paper awardee, “Densely Connected Convolutional Networks” is a minor extension on top of last year’s best paper (ResNet). Fortunately, the rest of the awards went to truly deserving works and I can recommend checking them out.Another work that I found quite interesting was on predicting future behavior of traffic participants employing deep inverse optimal control [3].Finally, a magical, nearly unbelievable work is [4] which uses a CNN to reconstruct a few hundred dimensional EEG signal recorded by showing human participants images of 40 ImageNet classes. The paper shows that classifiers trained on this feature representation (mimicking human cortical spikes just for 40 classes) perform en par with traditional CNNs trained on 1000 ImageNet classes! Not only that, the authors announced their accepted ICCV paper, where they invert EEG signals to images that the human is looking at, employing GANs. They are able to reconstruct what you are seeing with your eyes just using EEG signals from your brain!!! Kaboom! That just blew away my mind. I have subscribed to the authors’ Google Scholar profiles and eagerly look forward to that ICCV paper becoming available.2. Return of domain knowledgeOne major shift in the tide that I felt this CVPR, is towards incorporating explicit domain knowledge into otherwise data-driven models. It seems that the community spent the last couple of years exploring rather generic CNNs/LSTMs/GANs and has started seeing a plateau in performances.There are two ways by which model-driven ideas are being revisited: with novel layers or losses and by exploiting auxiliary training objectivesMy favorite paper at the conference, that I hadn’t even seen before is called “Learning shape abstractions by assembling volumetric primitives” [5]. It revives decades old ideas that represented 3D objects using generalized cylinders and geons, but in a CNN context. The method learns to encode an object class using a small set of oriented cuboids, employing an autoencoder with a constrained output representation for the task. I have been thinking hard about the problem of representing 3D objects using primitives myself, and couldn’t come up with such an elegant (almost) unsupervised solution – nor have I seen anything like this in recent literature.Another interesting work is called “Deep Kinematic Pose Regression” [6] which proposes a novel layer and loss to enforce that human body pose estimates emanating within a CNN obey real kinematic constraints.3D face reconstruction is performed in [7] from a single image, by introducing a proper graphics renderer as a layer inside a CNN.Since part-level reasoning is even more important than global object-level reasoning for fine-grained categorization, [9] leverages an attention module sandwiched between a coarse-scale and fine-scale CNN layers as in [7].Apart from exotic layers and losses, the second route to injecting domain knowledge into CNNs is through auxiliary supervision. Our own paper [10] shows how a CNN can be trained entirely on (photo unrealistic!) synthetic data – without pretraining or fine-tuning on real images - by supervising the network along depth by multiple relevant concepts in order. As opposed to supervising with a single task, or even multiple tasks at the output node, this strategy avoids overfitting, and brings to bear the unique richness of labels and modeling of occlusions that 3D CAD data provides. Ofcourse this observation is not limited to us: at a workshop Raia Hadsell talked about some exciting work on deep reinforcement learning applied to agents navigating 3D mazes taking in RGB images, where training the network to predict depth as a side-task helps even more than providing depth as an input!At a workshop, Fei-Fei Li described her recent attempts at reviving GOFAI [17] ideas from the days of Winograd, to improve visual-language modeling. While her group’s initial attempts at image captioning type of problems were purely data-driven, she doesn’t believe language problems can be solved using data alone, irrespective of the amount of training data available. Together with learnable components some degree of explicit symbolic modeling will be indispensable!3. Differentiable modules:While there were plenty of papers trying to introduce domain knowledge into data-driven approaches, there were others trying to introduce data into domain - proposing to replace still remaining explicit algorithmic blocks in computer vision pipelines with differentiable modules.Learning non-maximum suppression [11] reasons about pairs of object detection hypotheses emerging from a generic object detector and rescores them to ensure that there will only be a single bounding box predicted per object when the detections are thresholded by score.Deep Network Flow for Multi-Object Tracking [8] incorporates a linear program as a layer in a CNN for the task of multi-object tracking.Leonidas Guibas’s group presented a couple of papers on directly applying CNNs to point clouds, avoiding the need to convert them to a more structured voxel grid representation first [13].3D keypoint descriptors are learnt in [14] like many recent works for 2D, but I found the application to intra-class 3D CAD model alignment to be particularly exciting. This capability may enable utilization of 3D CAD models for object recognition, which is something that I have built my career around, in a completely unsupervised manner! Note to self: actually [5] may also be relevant for this.4. Dataset papersSince the revival of deep neural networks, every other paper at the last few CVPRs has been proposing a new dataset. This trend continued at this meeting too.The one dataset paper that I really liked had released a synthetic 3D dataset comprising of 40K houses with finely labeled objects, at the level of CAD model alignments [15]. This is going to be of major help to researchers working in 3D scene understanding, going beyond bounding box level object recognition problems that are only relevant for web applications but useless for robotics.Industrial ExhibitsThe tradeshow at CVPR has grown exponentially over the past couple of years. This year we had dozens of startups demoing really cutting edge examples of productized computer vision. In particular, it seems that Chinese startups can make any vision algorithm work in real-time on palm sized smart cameras that burn only a few watts! My previous job was in the SF Bay area, and I got adequate opportunity to visit and interact with vision startups there [thanks to Quora!], but I have a strong feeling that Shenzen is beginning to kick Silicon Valley’s ass when it comes to vision/learning innovation.One of these startups is Vion Technologies Inc. founded by SRI veterans, who were demoing half a dozen different applications in 100% edge-based visual surveillance. One cool new deployment they have made is at a school where their tiny device recognizes faces of children leaving and entering the campus and sends SMS messages to their parents. I have not seen smart cameras with these kinds of capabilities before.A startup called PIXM from New York (I think), was demoing their computer vision based solution to prevent phishing on the internet. I found the idea and execution to be incredibly convincing. I keep needing to tell my less technically sophisticated family and friends to avoid clicking on suspicious looking links, and yet even graduate degree holders in CS I know have entered all their details on a fake Amazon page! I feel dumb that I never imagined computer vision had a part to play in solving this.My team mates at Microsoft revealed our new neural network chip specifically targeting edge intelligence, as well as a research mode for HoloLens, which means that researchers will now be able to directly access the sensor streams inside of HoloLens. As a postdoc exploring semantic SLAM a few years ago, I was frustrated at the sheer amount of engineering effort it took to get precise camera tracking to work, before I could start incorporating visual recognition into the pipeline. So, I have no doubt in my mind that HoloLens’ research mode will have an even bigger impact on robotic vision research than Kinect.I wonder if soon CVPR will replace CES for vision and learning – with CEOs unveiling their latest products here! Speaking of CEOs, I spotted Nvidia’s Jensen Huang roaming around this CVPR as well, just like last year. I know that a few friends of mine received direct recruiting emails from him after last year’s meeting!! This sounds like a great way to ensure all the upcoming stars land at your company.In short, if you want to hire top talent in vision and learning, or you are a startup looking for business, CVPR is the place to be!Workshops & KeynotesThe best talk I heard at CVPR by far was the keynote by James DiCarlo. He talked about how advances in deep neural networks are helping explain what happens in the human visual cortex. If you weren’t there, I can strongly recommend watching the talk when it gets posted online. ** Edit: Now available at [20], thanks Kelvin Sudani for the link.I loved the workshop on “Visual Understanding Across Modalities” hosted by Ali Farhadi of Allen Institute for AI (A2I). A2I organizes three vision plus AI challenges that are all ahead of the curve. I specially love the Thor challenge [18], which involves robot perception and motion planning in realistic indoor environments while bypassing the manipulation problem. As opposed to deep RL driven navigation and planning work being done at DeepMind in very artificial maze like environments, I found Thor to be a lot more convincing.Another nice talk by Jianxiong Xiao at the “Deep learning for robots” workshop included pausing Tesla’s inner-city self-driving video and pointing out how crappy their object detector is: highlighting that we should have little faith in demo videos! Anybody can record a hundred demo experiments, and share the one instance where their system works!While many people in the robotics community are leveraging deep reinforcement learning to teach robots by demonstration, Ashutosh Saxena mentioned some of his work [16] on jointly modeling language, perception and planning that sounded amazing.I have long wondered [19] why we can’t just simulate physical contacts for robotic manipulation or locomotion (think legged robots), instead of needing to build real robots. Jitendra Malik finally answered my question, remarking that the challenge is not in modeling the physics behind these interactions, which is quite simple. The real problem is that we don’t yet have detailed catalogs of material properties like friction, elasticity which are needed to make these models work. He pointed out that the graphics community had to go through a similar elaborate exercise over the past few decades modeling how light interacts with various materials, before they managed to succeed in delivering realism in their works. Perhaps a brave young PI can take up the challenge and build her career around this problem? I know it will be a huge boost to the broader AI community.Most importantly, Honolulu is an amazing amazing place to visit. I feel fortunate that we as a community are commercially relevant enough that we don’t have to hold our conferences in Omaha! It was very clever of the organizers to leave half a day free in the middle of the conference, so people could do a bit of tourism. I didn’t get to visit any of the other Hawaiian islands, but did go to a couple of beaches, Pearl Harbor sites, as well as for a hike on the Three Peak mountains trail. This trail passes through thick forest, and on my way back, I got separated from my group and lost in the forest and ended up calling 911 after spending a couple of hours searching for the exit (there were fences and abandoned cabins in the middle of the forest, and mosquitoes, lots of them)!! Hats off to the amazing fire department staff who immediately came to my rescue. Request to HERE Maps people: Can you please add a feature to show latitude and longitude in your app as well?That’s all folks! See you next year in Salt Lake City.[0] Hollywood in Homes: Crowdsourcing Data Collection for Activity Understanding.https://arxiv.org/pdf/1604.01753.pdf[1] Counting Everyday Objects in Everyday Scenes https://arxiv.org/pdf/1604.03505.pdf[2] Image-to-Image Translation with Conditional Adversarial Networks.https://arxiv.org/pdf/1611.07004.pdf[3] DESIRE: Distant Future Prediction in Dynamic Scenes with Interacting Agents.https://arxiv.org/pdf/1704.04394.pdf[4] Deep Learning Human Mind for Automated Visual Classificationhttps://arxiv.org/pdf/1609.00344.pdf[5] Learning Shape Abstractions by Assembling Volumetric Primitiveshttps://arxiv.org/pdf/1612.00404[6] Deep Kinematic Pose Regressionhttps://arxiv.org/pdf/1609.05317.pdf[7] Learning Detailed Face Reconstruction from a Single Imagehttps://arxiv.org/pdf/1611.05053.pdf[8] Deep Network Flow for Multi-Object Trackinghttps://arxiv.org/pdf/1706.08482.pdf[9] Look Closer to See Better: Recurrent Attention Convolutional Neural Network for Fine-grained Image Recognitionhttp://openaccess.thecvf.com/content_cvpr_2017/papers/Fu_Look_Closer_to_CVPR_2017_paper.pdf[10] Deep Supervision with Shape Concepts for Occlusion-Aware 3D Object Parsinghttps://arxiv.org/pdf/1612.02699.pdf[11] Learning non-maximum suppressionhttps://arxiv.org/pdf/1705.02950[13] PointNet: Deep Learning on Point Sets for 3D Classification and Segmentationhttps://arxiv.org/pdf/1612.00593.pdf[14] 3DMatch: Learning Local Geometric Descriptors from RGB-D Reconstructions3DMatch: Learning Local Geometric Descriptors from RGB-D Reconstructions[15] Semantic scene completion from a single depth imageSemantic Scene Completion[16] Deep multimodal embedding: Manipulating novel objects with point-clouds, language and trajectorieshttps://arxiv.org/pdf/1509.07831.pdf[17] Visual Genome: Connecting Language and Vision Using Crowdsourced Dense Image Annotationshttps://link.springer.com/content/pdf/10.1007%2Fs11263-016-0981-7.pdf[18] THOR[19] http://www.zeeshanzia.com/pdf_files/proposal_NTNU.pdf[20]

I don't know for sure, but since you dare to ask, here are a few thoughts.There's no such thing as the perfect screwdriver. It depends what you are trying to do, stir paint, or eat a coconut. Different screwdrivers have different uses.So, what would your 'effective' look like?Are your students expected to speak effectively, or get good marks? If the assessment regime is dodgy, these will not be the same. It never will be exactly (absolute pontification!).Are the students expected to ... (do they assess oral English with pencil and paper testing, which is pretty common in school systems?) ... give speeches that impress their families, or to communicate?They are rarely the same. 'Communicate' carries the implication of having a useful impact on the listener (we rarely 'want' the impact ... boredom, impatience, superficial entertainment). I like a definition from NLP, to the effect that 'communication' leaves the recipient different, other than bored.Are they expected to communicate like rappers from the ghetto, or Harvard Business School equivalents? Not the same. Do your students talk in text-speak?To answer your question effectively, it depends upon what your employers (or just you if you do not intend to stay long) think that effective oral English is.Will your best graduates reflect your program in their oral oddities? Our English tells Sherlock Holmes where we got it, whoever we are.'Here's your fla why', when they serve people coffee, because they have been told that native speakers hardly vocalize the final consonants?So. Get clear about the goals that matter. I think.And, what do you have to fall back on?My contestable solutions to a similar problem, some years ago, just after my eleventy-eleventh birthday:20 something year-old college students who had had maybe two English classes per week for the whole of their schooling. Mostly conducted in their mother tongue, being told about English, plus selected exercises. Teacher typically sitting at the front. Students often falling asleep. A presumption that if you put in the hours, that was what good character amounted to, even if your progress was zilch, 'I tried!', so some arrived in the classroom at 7am, and left at 10pm, intermittently transcribing and reading their textbooks, and falling asleep, and sooner or later jumping out of an upper floor window.I had a slight introduction to language teaching, a rather sad attempt with excellent trainers, to master CELTA, plus a million years of teacher training (called 'Teacher Education' for branding purposes) presentations and discussions. Plus quite a bit of experience in group process, trance, and Tantric eating and resting.Plus maybe 5 years in Toastmasters, eliminating my 'ums', and learning to throw together a beginning, middle, and end, at short notice. And eventually, to talk on an impromptu topic inside two minutes. That took a while.So, I can't really answer your question, because it depends entirely on your different context. Near enough is different. Including your leadership. Mine was very 'elusive'. More or less, 'just do your thing, and we'll see whether we like it.'And what would I know anyway?So given the desks in rows, with the podium at the front, I got in early, piled the desks against the back wall, and placed the chairs around the room, so that we sat in a big circle.'Huh'? Half of my mostly female (English Majors) cooperated with whatever training games we came up with; the other half wanted to go shopping. I manipulated the assessment regime so that they picked up tiddly marks for being five minutes early, and coming back from the loo.Every kind of game and task that you could imagine. Few set speeches: they could do that, sometimes without understanding a word.The greatest challenge emerged in throwing them into teams, A-B-C-A-B-C, etc., around the room. Mix them up. Then, 'A', you are the boy-friend, 'B', you are the girl-friend. 'C', you are her girl-friend.All the 'C's came outside with me for secret instructions: they were to steal the B/F. In social chat, all in English.About the time that the plot started to reveal itself, the action exploded into the mother tongue, quite dramatically, as and C started punching each other, so eventually I found it useful to carry a water pistol. With contrived permission, my job was to prowl, and squirt anyone who wasn't using English. Got away with it. Might not. (Didn't squirt them for assault. Just the way that they dd things.)So obviously that was all unwise, and would never work. It did there, and for a while.Each semester I found myself allocated to a more remote classroom, and offered more classes for extra pay. The Dean prowling the corridor, peering through the glass door. 'Shhhh'. Really?Classes passed quickly.I assessed them based on one-on-one interviews ... a big job ... it was oral English.There was little to lose. Most classes in most subjects do not assess that which is the reason for the class. How many management courses assess management? They assess knowledge of ...? Different. Hence graduates are often unimpressive. They are the winners of the college game,How many clergy programs effectively assess integrity? Or wisdom? Would a wise person become a clergyperson?I decided that I was going to spend my time lifting the confidence and oral communication capability of my students, come what may. Some enjoyed it. Some threw themselves into it. Some didn't.My program was limited by my skills, competence, and imagination, and had to find its place in that particular context. And, really, what do I know? :-)(Obviously had an ego-centric time :-)Wouldn't work elsewhere.

I graduated with an IB Diploma in 2016. My overall thoughts are a mixed bag really, mostly positive but I’m going to write this as objectively because I know there are some that will ask “Well what about AP or A levels?”So. Here’s my take on it. Beware, this answer is going to be a long one because I want to emphasize on the pure amount of work there is to do and how I felt about it:Yes it’s tough but is it worth it?Yes the IB Diploma is rigorous and challenging. There’s no hiding it. It’s touted by IB educators as being tough and a real challenge for high school students. Is this a bad thing?Depends on who you are.Remember, these are the final 2 years of high school right before university. University is a whole new level of stress, anxiety, and lack of sleep. I like to think of IB as the appetizer before the main course. I’m personally glad I got to experience what I did in IB.Of course I moaned and complained the whole way. All IB students complain about how difficult it is but that’s just part of the package. IB really did teach me things that I never would have learned at A levels or AP. I’m certain about that.Now was it worth it? Sure it was. The skills I learned as an IB student and the values of things like time management and proper citation in essays is something I’ll never need to worry about because I was exposed to a program that emphasized both.Now all that being said, will universities look more favorably on students with an IB Diploma high school certification? Personally, I think it’s unlikely. I can’t imagine professional teaching institutions favoring any type of certification over the other.Is it recognized? Absolutely. It’s recognized by more and more universities each year. You won’t have a problem applying to the UK, US, or Netherlands (which is where I applied).Essentially it depends on what you consider important. Experience or convenience. If convenience is what you prefer, then AP and A levels are a no brainer. You take 3 or 4 subjects, study only for those subjects, take an exam, get graded, and then send those grades off to university.On the other hand, you’ve got IB where you need to do a mandatory 6 subjects (unless you’re doing the IB Certificate and not Diploma) as well as a 4000 word Extended Essay, 1600 word Theory of Knowledge Essay, 10 to 30 minute Theory of Knowledge presentation, CAS (Creativity, Action, and Service) which is basically voluntary workBottom line? Universities will see A levels and AP as being just as good as the IB Diploma. So if that’s you then IB may not be for you.If it’s experience and the “life long learning” that IB advertises then read on.Practical Experiences that helped develop me as a studentIt’s not secret that on top of 6 subjects, you’ve got a helluva lot to do. On top of studying for exams at the end of the 2 year course you’ve got:A 4000 word Extended EssayA 1200 - 1600 word Theory of Knowledge EssayInternal AssessmentsThese are mini research essays for your humanities and science subjects like Economics, Business, Physics, Biology etc. that can range from 750 words (Economics) to 1500 - 2000 words (Business and Physics)Written tasksFor your language subjects like English, French, Spanish. The lengths of these tasks varies depending on whether you do an A or B language and also whether you do Higher Level or Standard Level.Further Oral Activities (FOAs)These are for the language subjects as well. These are essentially graded presentations you do by yourself or in a group on a part of the specific language courseInternal Oral Commentaries (IOCs)These are solo presentations that are recorded to be marked externally by IB Examiners. These are usually analyses on literary works that the students have done in their language subjects.Creativity, Action, and Service (CAS)This is voluntary work or sporting activities that students need to take the initiative to do (because if you don’t pass CAS then you don’t pass IB).They can vary from simple actions like taking part in a walk for breast cancer or more complex ‘CAS projects’ which need to be student organized events/activities.Now if you’ve read all that and feel overwhelmed, welcome to the IB club. Those are what a student doing IB would be expected to do on top of studying for exams and just dealing with everyday life.But notice how a lot of those include essay writing as well as presenting?Exactly.By the time I was done with IB, I was so well versed in writing a properly cited essay that it wasn’t a burden at all.As a matter of fact, in my first three months of university, we were given the task of creating a report as a group. A big issue that came up with many other groups was the lack of citation included.All groups except mine.Citing sources had already been drilled into my mind by IB. For this, I thank IB. Citing sources will never be a challenge ever again.As a counter-argument, some universities use different citing methods. The one I was taught at IB was APA citation. Friends of mine have said that they’ve needed to use MLA as a format of citing. Personally, I don’t think that’s an issue.Why? Well because once you’ve got the skills IB has made you teach yourself, adapting to new ways of citing isn’t a big deal because you’re already experienced with citing in general. It’s not a big deal to switch formats if you know the general concept inside out.As for the presentations, this is down to the student. I was personally very good at speaking and presenting so I was essentially refining how I presented as opposed to learning how to do it.What’s my verdict in this case?I think IB really got me ready for university in this case. Not only was a forced to manage my time and efforts as far as commitments to studying and coursework went, but I was taught the importance of citing and the negative impacts of plagiarizing.The dreaded Extended Essay and Theory of Knowledge EssaysI would say that I learned A LOT from just these two aspects of IB alone. The Extended Essay was a grueling process, stressful, and I spent many all nighters on it.Nowadays I just scoff at the idea that some of my peers shudder at the idea of a 4000 word essay. Been there, done that. Move on.The EE not only taught me how to cite, it also taught me to be independent. To do my own research. I did a science based EE so that meant I needed to come up with my OWN experiment from which I needed to get successful results from that would SUPPORT my essay.For that, I once again tip my hat off to IB. I was assigned a supervisor for EE, as all students are, but we were told specifically that there was to be NO spoon feeding (a common theme in IB).I was able to create a 4000 word essay that did way better in terms of a grade than I ever imagined it could.Ask yourselves if an A level or AP student could create a 4000 essay to the calibre of an IB student. I think the answer is pretty clear. We were taught something practical. Something universities would soon be testing us on FOR OUR GRADES.The TOK essay is a whole different story. TOK is the reason I’m able to write this response in the (hopefully) objective and thoughtful manner that it is. It’s the reason I’m able to sit down with someone and debate a certain topic.It’s the reason I can think outside the box and ask essential questions.Now for me, that was important. Had I done AP or A levels, there’d be no way in hell I could be as much of a “thinker” as I am now. Once again, IB takes the win for this one. A levels and AP would’ve just taught me that facts matter because they’re graded. However, that’s just me personally.So on the whole for this point, I really can’t thank IB enough for teaching me the practical stuff that would equip me for university. I realized quite quickly that I would be able to cope with university just because I’d been through IB. Again, this is just my opinion.Creativity, Action, Service. Not too fond of this one to be honest.As the title suggests, I have to be honest. Out of everything IB taught, CAS didn’t really stick. That’s just me though. I haven’t gone off to join marathons or walks or any other events. Haven’t planned a single event either.CAS was merely something that needed to be done. It was a task that we had to complete just to pass IB. I’m naturally very sporty and I go to the gym once in a while but I don’t find myself really volunteering for anything.The way CAS was done at my school, you could throw a piece of litter into the trash and that would count as an activity you did on your CAS journal. Need I say more?So yeah. Very brief but CAS really didn’t do much for me.Social LifeThis is an interesting one because it may be down to pure circumstance or location. My school was in Dubai and during IB, I formed some very close friendships with a lot of the folks at school, all of which I am still in contact with.I had never had such an experience. I moved around a few schools because my parents weren’t satisfied with the education of the others but as soon as I started IB, I started making some real, life-long friends.Like I said, this may be because I became more social or because of the school or the fact that it was Dubai. Whatever the case, I’m sure most IB students can agree that the crap everyone goes through together brings everyone closer.IB students know the pain. They can relate to each other. There are groups and memes all over the internet that are inside jokes to all IB students as a result of IB.For this, I truly think IB did me some solid favors. I found some amazing friends, all of whom I can relate to because we did the same certification.So for me, I really think IB helped me have an awesome social life along side the rigorous studying.The stress, anxiety, mental, and emotional rollercoasterIB inflicts a certain amount of stress on all IB students. Ironically this is a good way of telling who’s more organized.For me, the amount of stress, anxiety and every other negative emotion I felt in IB could have been alleviated by time management, something I was never good at in IB but have gotten better at since because the IB experience taught me it’s value.It’s quite a long two years. There are all nighters (a sure sign of bad time management) and plenty of coffee. It was tough. No doubt about it but let’s be real. Life is tough.Any high school student going on to university will tell you that university is a whole different level of tough.And then what happens? You get a job of course. Whoever heard of a job that was easy and enjoyable? Unless that job is your passion and you love it, guess what? It’s likely going to be VERY tough.A levels and AP may be easier and less stressful. I’ll never know. What I do know is that IB gets it’s students into a mental state of preparation. This can obviously be argues by others but to me, IB was just a crack in the window compared the fully open windows that is university life.It makes me really want to go back to IB whenever I’m at university. I find myself wishing for the days of IB, something current IB students would be shocked about.If the challenges of IB are a turn off for some then fair enough. My personal belief from my experience at university so far and from IB is that it doesn’t get easier. Life will have it’s challenges but luckily for students, IB is an optional challenge.You can either take the IB Diploma and get a sneak peek into the real world of all nighters, stress, anxiety, emotional craziness or you can relax and just make sure you’ve memorized all the facts you need for AP and A levels.The choice is yours. Personally, I’m glad I chose IB.The Verdict (at last)This has been a long answer but I know there are some reading this who really may be torn between which certification to go along with (A levels, AP, or IB).To me, IB was a once in a lifetime opportunity. Call me insane but if I had the chance to go back, I probably would just because it really smacked me into shape. I can say wholeheartedly that AP and A levels would never have done what IB did for me (or what I did for it ironically enough).I legitimately and objectively learned valuable life skills, learned how to think critically and be curious, met a wonderful social group of friends, including teachers, and made it through it all with a 37 which was more than enough to get me accepted to the universities of my choosing in the Netherlands.No IB is not perfect. No it does not teach you knowledge that you need for university but why should it? Do AP and A levels teach you information you need for university? I don’t think so but then again I’ve only done IB.I can vouch for IB because it taught me everything I said up above in the preceding paragraph. Is it stressful? Yes. Is it difficult? At times yes. Is it worth it?Depends on who you are but for me, IB was possibly the best way I could have possibly said goodbye to my high school days and hello to my university life.