Aug 24, 2017

Everyday automation through bioinspiration

0 comments

Edited: Aug 24, 2017

Another mental model to adopt when looking for transformation through AI techniques is the exploration of human processes and the adaption of them into a digitised form. The processes can be varied; the contextualising of input feeds, the conclusion reaching cognition of experience or unblinking flawless attention to changing events.

 

The discovery process is somewhere between anthropology and biomechanics; taking inspiration from the biologically improvised and optimising, perfecting, automating.

 

 

 

New Posts
  • Short answer: - No. Slightly longer answer: - Not yet. Questions you may be asking yourself Should I be worried? - Again, not yet. Then what are you talking about? - Let me explain: A few days ago, I saw this video circulating Twitter highlighting a piece of software from https://teleporthq.io/ , which shows a person drawing the design for a web-page on a whiteboard. There is a camera pointed at the board and a screen to the side showing the same design as the one being drawn; as the person draws more, the on-screen design develops alongside it, with placeholder text and images reflecting what had been drawn. From what I can tell in the video, the output of the image recognition is a React JS file. Straight from the off this seems like a revolutionary breakthrough and the thing that’ll be replacing everyone’s jobs in the near, however, on further consideration I’m not quite sure that’s the case. This seems like an incredibly useful tool, yet, not quite for creating full blown websites (for reasons I’ll speak about shortly) but possibly absolutely outstanding for smaller tasks. Due to its ease of use and rapid production of visuals, it could potentially be a great tool for developing prototypes – it would not only allow for non-techies to show their own ideas created before their eyes simply by just drawing a few shapes on a board, but it will also produce something which will be able to be used to some limited extent. This could have a huge positive impact on understanding usability prior to building sites by seeing how the user will want to use it and changing the design within seconds. As previously mentioned, I personally don’t feel (at least from the limited amount I’ve seen) that this software would actually be that useful for creating an entire website. There are a number of reasons for this. First off, functionality. From my understanding this software is aiming to make digital user interface designs codeless. Yet there doesn’t appear to be any way to even add actions to buttons or add links between pages; something that should be a given seeing as it’s producing React JS; these would have to added manually into the code – something made all the more daunting by my next point. Maintainability. Someone is going to have to maintain all these pages. From the video, you can see that the view is created by a single render of a whole bunch of components inside a <Page> wrapper – for more complex designs this could become a monolith of a file which someone would have to manually go through to add classes, ids or alternate functionality. And whilst I’m mentioning classes and ids, in this video there is nothing shown on how the CSS is complied, or what the final CSS file looks like. In another video from their blog page, it shows that the CSS is actually complied as a react constant within the produced React file. It’s extremely verbose and uses absolute positioning almost everywhere which frankly is pretty bad practice and again would make maintenance a real pain. Furthermore, what happens when an update is needed on the page? If this is meant to be codeless, would you have to redraw out the whole design again and hope that it infers that the design is the same as it was previously? And on top of this, would you then have to go through and make all the manually added functionality again, as it’s essentially a rebuild? One of the largest issues I feel may come of people using this however, is not to do with the system, but actually in fact to do with human laziness. If people use this as a low fidelity prototyping tool and then customise it over time in order to make it look, or work, a little more like they expect it to, then it could slowly mutate into a horrific bastardisation of part auto-generated, part customised code in a single behemoth of a file. But as I say, this is not an issue with the program, but to do with human nature. A few other things to think about would be code conciseness. As I previously said, the output of this is overly verbose on first glance. But I’d be intrigued to know how it would react (pun intended) to the creation and/or grouping of common parts into reusable components if there are sections of the pages which are similar in nature; especially if they’re on separate pages. However, these flaws could all just be due to the infancy of this technology and I look forward to seeing how this kind of software grows. Nevertheless, it’s showing something promising and even if it goes no further, it still provides an amazing tool for simple prototyping.
  • Website creation has lately been changing in order to appeal to the rapidly growing market of those not particularly skilled in the area, who are wanting to build websites. CMS went some way to solve this as people could completely remove themselves from the code-based technicalities of web development, such as WordPress. Despite this many people still struggled with the whole design side of website creation. In an attempt to solve this, CMS based around templating which restrict the freedoms of design (in order to ensure websites are usable and consumer friendly) such as Wix and Squarespace were created. However, all of these CMS are still not entirely self-explanatory, with problems like confusion over naming of menus (due to menus being created by those who are technologically minded). This is where AI is starting to creep in. A number of companies are attempting to automate both the design and development of the website through AI systems. One of the most prominent at the moment is Wix’s own ADI ( Artificial Design Intelligence ). With this system, you simply give it your company name, the type of company you are and a few words about what your company does. Wix then returns with a complete site, sub-pages and all – based on what it thinks you should have on your site – ready for you to drop in all of your content. You may be thinking, but what happens if you don’t like it? Well, you can either restart and roll the dice again, or use a range of menus which let you change very broad things on the site; colour palettes and the kind. But let’s take a look at the example result they give: � � It’s an appealing site, I’ll agree with that. However, Wix say “no two sites ever looking the same”, yet, to me, there doesn’t appear to be anything special about this site. But reading into this a bit further, I may have found why. “…choosing from billions of high-quality, stunning combinations and possibilities...” Wix’s ADI seems to only look through its own templates, taking parts of them and merging them together into another new template; hence why there can be billions of combinations without any fear of exact repetition. But is this really the best we can do? I took a look at another AI designer, called Firedrop , to find out. Similar to Wix, Firedrop attempts to make design easier by minimising the learning curve needed to create a website. However, Firedrop does this by using a chatbot for an interface which starts in the exact same way as Wix – asking for your company name, type and description. After that all you need to do is type what you want to do and everything else is done for you. So again, let’s take a look at the 4 example outputs: � � Now, these are nice websites. But… they’re all so similar. And even comparing these designs with that from Wix again there’s huge similarities. They all have a large image at the top, followed by sections of text with a large image to the side. I’m not saying that these are bad AIs by any means. It just seems that they aren’t necessarily as impressive as you may first think. It would make sense that the systems work by analysing websites with similar titles, or descriptions and pull together a template that has features used on the analysed sites. However, this leaves the outputs as nothing more than an “average”. From my point of view these AI created sites aren’t unique and bespoke, but a little more generic than you would hope. Just to reiterate, I’m not saying that these systems aren’t good, I just personally think they are far from being widely adopted as in their current state, they’re more of a gimmick than something that’s actually useful. If these sites continued to churn out site after site, I would image at some point, they would begin to analyse themselves; potentially causing averages of averages. If this did happen and these AIs were widely adopted, then it would indirectly cause many issues. I feel fewer people would choose to try and create bespoke sites by hand, or develop new software, JS plugins, CSS attributes, etc. Issues would arise by AIs needing to learn how to use these changed, or just omit them completely, causing web innovation to slow.­­ To conclude, I think that this is a good start for AI in web design, but there needs to be definite improvements to make the outputs less alike. Also, perhaps adding some kind of rating on how well the AI performed would improve the diversity throughout the sites, without having to redo the entire creation process. If you want to see a bit more about these two technologies check them out at the links below. Wix: https://www.wix.com/blog/2016/06/wix-artificial-design-intelligence/ Firedrop: https://firedrop.ai/
  • UK Government Department of Business, Energy & Industrial Energy has issued its review of AI and its impact on the economy, here is a summary; Increased use of Artificial Intelligence (AI) can bring major social and economic benefits to the UK. With AI, computers can analyse and learn from information at higher accuracy and speed than humans can. AI offers massive gains in efficiency and performance to most or all industry sectors, from drug discovery to logistics. AI is software that can be integrated into existing processes, improving them, scaling them, and reducing their costs, by making or suggesting more accurate decisions through better use of information. It has been estimated that AI could add an additional USD $814 billion (£630bn) to the UK economy by 2035, increasing the annual growth rate of GVA from 2.5 to 3.9%. Our vision is for the UK to become the best place in the world for businesses developing and deploying AI to start, grow and thrive, to realise all the benefits the technology offers. The pioneering British computer scientist Alan Turing is widely regarded as launching and inspiring much of the development of AI. While other countries and international companies are investing heavily in AI development, the UK is still regarded as a centre of expertise, for the present at least. This report recommends that more is done to build on Turing’s legacy to ensure the UK remains among the leaders in AI. Key factors have combined to increase the capability of AI in recent years, in particular: New and larger volumes of data Supply of experts with the specific high level skills Availability of increasingly powerful computing capacity. The barriers to achieving performance have fallen significantly, and continue to fall. To continue developing and applying AI, the UK will need to increase ease of access to data in a wider range of sectors. This Review recommends: Development of data trusts, to improve trust and ease around sharing data Making more research data machine readable Supporting text and data mining as a standard and essential tool for research. Skilled experts are needed to develop AI, and they are in short supply. To develop more AI, the UK will need a larger workforce with deep AI expertise, and more development of lower level skills to work with AI. This Review recommends: An industry-funded Masters programme in AI Market research to develop conversion courses in AI that meet employers’ needs 200 more PhD places in AI at leading UK universities, attracting candidates from diverse backgrounds and from around the world. Credit-bearing AI online courses and continuing professional development leading to MScs Greater diversity in the AI workforce An international AI Fellowship Programme for the UK. The UK has an exceptional record in key AI research. Growing the UK’s AI capability into the future will involve building on this with more research on AI in different application areas, and coordinating research capabilities. This Review recommends: The Alan Turing Institute should become the national institute for artificial intelligence and data science Universities should promote standardisation in transfer of IP Computing capacity for AI research should be coordinated and negotiated. Increasing uptake of AI means increasing demand as well as supply through a better understanding of what AI can do and where it could be applied. This review recommends: An AI Council to promote growth and coordination in the sector Guidance on how to explain decisions and processes enabled by AI Support for export and inward investment Guidance on successfully applying AI to drive improvements in industry A programme to support public sector use of AI Funded challenges around data held by public organisations. Our work has indicated that action in these areas could deliver a step-change improvement in growth of UK AI. This report makes the 18 recommendations listed in full below, which describe how Government, industry and academia should work together to keep the UK among the world leaders in AI.