[fusion_builder_container type=”flex” hundred_percent=”no” equal_height_columns=”no” menu_anchor=”” hide_on_mobile=”small-visibility,medium-visibility,large-visibility” class=”” id=”” background_color=”” background_image=”” background_position=”center center” background_repeat=”no-repeat” fade=”no” background_parallax=”none” parallax_speed=”0.3″ video_mp4=”” video_webm=”” video_ogv=”” video_url=”” video_aspect_ratio=”16:9″ video_loop=”yes” video_mute=”yes” overlay_color=”” video_preview_image=”” border_color=”” border_style=”solid” padding_top=”” padding_bottom=”” padding_left=”” padding_right=””][fusion_builder_row][fusion_builder_column type=”1_1″ layout=”1_1″ background_position=”left top” background_color=”” border_color=”” border_style=”solid” border_position=”all” spacing=”yes” background_image=”” background_repeat=”no-repeat” padding_top=”” padding_right=”” padding_bottom=”” padding_left=”” margin_top=”0px” margin_bottom=”0px” class=”” id=”” animation_type=”” animation_speed=”0.3″ animation_direction=”left” hide_on_mobile=”small-visibility,medium-visibility,large-visibility” center_content=”no” last=”true” min_height=”” hover_type=”none” link=”” border_sizes_top=”” border_sizes_bottom=”” border_sizes_left=”” border_sizes_right=”” first=”true”][fusion_text columns=”” column_min_width=”” column_spacing=”” rule_style=”default” rule_size=”” rule_color=”” content_alignment_medium=”” content_alignment_small=”” content_alignment=”” hide_on_mobile=”small-visibility,medium-visibility,large-visibility” sticky_display=”normal,sticky” class=”” id=”” margin_top=”” margin_right=”” margin_bottom=”” margin_left=”” font_size=”” fusion_font_family_text_font=”” fusion_font_variant_text_font=”” line_height=”” letter_spacing=”” text_color=”” animation_type=”” animation_direction=”left” animation_speed=”0.3″ animation_offset=””]
Artificial intelligence (AI) surrounds us all. AI, or more correctly termed machine learning, can be found in many different applications in your home, car, and workplace. Our relationship with machine learning applications started when we began to use voice-operated smartphone assistants like Apple’s Siri or Google’s Voice but voice recognition is just one example of a machine learning application; others include vision processing and object recognition.
Okay, Google
With the wide range of industrial and electronics sector clients Publitek has, machine learning is a frequently covered topic. Over the past ten or so years, we’ve seen the use of machine learning (ML) neural networks, once restricted to compute-intensive data centers, emerge to bring intelligent control to the edge of industrial applications. It isn’t always apparent, but many of the voice recognition applications we interact with rely on constant cloud connectivity to process our speech. You have probably encountered that when asking your smartphone assistant a question when you’re in an area of poor cellular reception. Voice-operated intelligent home devices need to constantly listen to your room’s ambient sounds for a command trigger word or phrase, for example, ‘Okay, Google.’ Once detected, the device records or streams an audio file to the cloud data center to process and identify what action you desire. The command word or phrase is the only part of the voice processing that is inferred locally.
Machine learning industrial applications
Regular technical content topics for our semiconductor clients cover inference – the term used to describe how ML determines the likely word or object – at the edge and TinyML. TinyML perfectly describes how voice, event, or object recognition is achieved on low power, low compute capability microcontrollers without internet connectivity.
At Publitek, we often create content about many aspects of machine learning developments. For example, a semiconductor client may offer a low-power microcontroller that, coupled with a vibration sensor such as an accelerometer, creates a vibration detection sensor for use with an electric motor supplied by one of our industrial clients. A sensor like this can detect a change, either sudden or gradual, in the motor’s vibration signature. It could indicate a warning to maintenance staff that an investigation and corrective action are required. This is a perfect example of the application of TinyML. No high bandwidth, always-on cloud connectivity is required, and the sensor can be battery-powered.
Can AI be used to write content?
The increasing abilities of AI and machine learning started to come to the attention of media companies several years ago. Keen to benefit from the economies that AI offers, media companies, particularly newspapers and news websites, have made their own headlines by announcing that they are replacing journalists with AI.
So, can AI write? The answer is definitely yes, but the real question is whether it can write well and whether the written copy makes sense and satisfies the reader? The answer to the supplementary question is not as convincing.
The OpenAI initiative has been working on several text creation projects since 2015 and has recently announced the latest version of its natural language generation (NLG) model GPT2. GPT2 and the associated NLG techniques are already widely used to generate automated text for dashboards, social media posts, and client portfolio updates. Companies such as the Associated Press and the Press Association use AI to write tens of thousands of reports each year.
Writing a short sports report is however totally different from writing a technical article or this blog. Several websites are now offering the opportunity to trial AI writing, for example, AI-Writer.
The reality of AI-based content generation
As it stands today, the results achieved from these AI-driven sites lack a professional writer’s structure and finesse. To experiment I tried three separate topics:
- [fusion_modal_text_link name=”article” class=”” id=””]Can AI write an article?[/fusion_modal_text_link]
- [fusion_modal_text_link name=”microcontroller” class=”” id=””]How to conserve power in an embedded microcontroller application[/fusion_modal_text_link]
- [fusion_modal_text_link name=”omelet” class=”” id=””]How do I cook an omelet?[/fusion_modal_text_link]
As you will see from the examples, the generated text is composed of largely unconnected paragraphs that have been researched from the web. The [fusion_modal_text_link name=”microcontroller” class=”” id=””]embedded microcontroller example[/fusion_modal_text_link] is probably unfair, but it highlights most of the key methods of power conservation. Unfortunately, they are not structured as a technical writer, writing for an engineering audience would. The sources used include examples not referenced anywhere else in the text – BikeComputer, for example. That said, it researched and wrote this in a matter of minutes and provided the reference sources. AI-Writer highlights that the text isn’t as good as a human writer can produce but also highlights the time-saving nature of idea generation.
Given the large numbers of organizations now investing in NLG applications, the likelihood is AI-written content will feature increasingly in our lives. After all, there are many content generation tasks that are routine, straightforward, and have a limited number of variables.
In the meantime, if you’d like to speak to a human being about content creation – technical or otherwise – then please [fusion_modal_text_link name=”contactForm” class=”” id=””]contact Publitek’s Strategic Content Team[/fusion_modal_text_link].
[/fusion_text][fusion_modal name=”article” title=”Can AI write an article?” size=”large” background=”” border_color=”” show_footer=”yes” class=”” id=””]
The most expensive and time-consuming part of making a model is to train it and update the weights and connections between the neurons in each layer. The neural network consists of 96 levels, and instead of mere trigrams it tracks sequences of 2,048 words. [Sources: 3]
GPT-3 needs some kind of prompt to start it. That’s how the Guardian article came about. Some people have told me that it is because the article was not written by a human author that we are more likely to get it right this time. [Sources: 3]
When Britain’s Guardian newspaper wrote an entire article on artificial intelligence in September 2020, commentators pointed out that it still relied on human editorial content. Austin Tanney, head of AI at Kaino, said: ‘We got three or four different items and put them together. The article was written by AI, but there was no human editorial in it. [Sources: 5]
After human evaluation of numerous algorithmic methods for identifying machines that write articles, [14] articles that contained obvious errors for humans were identified, and articles with time values were automatic identifiers of the fact that a person wrote the article. [Sources: 5]
Bots can be used to write blog articles, but the quality is not the same as with real people writing. Only time will tell whether computers can improve and replicate the creativity and sincerity of human bloggers. There is no known bot that has spawned a successful blog. An article written by a human is not a bot article. [Sources: 6]
Wired published an obituary of Marvin Minsky, an artificial intelligence pioneer who wrote bots. A Swede named Sverker Johansson has developed a computer program that can write Wikipedia articles. She has written a total of 2.7 million articles. That’s 85 percent of all articles on Wikipedia, more than any human ever wrote. [Sources: 6]
It only takes a few seconds to add a sentence. The AI selects one word at a time and considers the next. It’s far from perfect, and the prose is rough, with occasional non-sequitur, but the article becomes more coherent the longer it gets. To be clear, AI can write news articles that convince anyone who is not surprised to see them in a newspaper, but it cannot write real news articles with quotes and made-up statistics. [Sources: 2]
As you may have seen, a recent article in the Guardian was written by a robot. ‘I know my brain but I don’t have an emotional brain. My brain is capable of making rational, logical decisions. I can teach myself, I know how to read the internet and I can write a column. [Sources: 3]
It can generate complete text paragraphs that mimic any writing style. The best thing you can do is not tell the robot how to write it. OpenAI discovered a major problem with GPT-2. The public is not allowed access. Fearing abuse, OpenAI has blocked it for further research. [Sources: 7]
I sent Steven Pinker, the Harvard psycholinguist, a sample of GPT-2 “s prose. He wasn’t impressed, but the machines plausible gibberish explained it all. I put Pinker’s answer in a generator window, clicked on a mandala to add synthetic Pinker prose to the real thing, and asked people to guess the author’s language, but my instinct was to prevent the machine from taking power. I knew it couldn’t help me write this article, and I guessed there were a million ways to screw it up. [Sources: 1]
It cannot sustain thinking if it does not have to think. Deep learning works well when it comes to capturing angular patterns in our syntactic gymnastics, but it lacks a pre-coded basis of procedural knowledge, so that it cannot use its language skills for reason or conceptualization. An intelligent machine must execute this kind of thinking. [Sources: 1]
There’s a lot of logic in the game, and the sequence of events can be more than a bit confusing. PORR is lucky that AI can create them. The spin-off of an article from GPT-3 is known in the media as an article published after an earlier article had already been published. [Sources: 0]
None of the media covering the story cared. In fact, I think the points raised in the comments are more interesting than the article itself. The media and Porr himself highlighted a comment in which one user expressed his suspicion that the article was written by GPT-3, which was downplayed by others. [Sources: 0]
He never mentioned the word “fake” in his own account of events. He did not attempt to influence public opinion by misrepresenting events. When I looked at Porr’s blog, I didn’t see that the definition of fake was applied to him. [Sources: 0]
They did not provide the speech generator themselves, but the OpenAI team gave me access to it so I could check their results before writing the piece. In the next 15 seconds, OpenAI’s new language AI extended the article for me. And of course I did the obvious: I asked the AI to finish the article itself. [Sources: 2]
I am grateful that the letter I read was in my own bubble. I did not interact with people who did not share my beliefs. However, I wasn’t sure how the article would be received by people from different backgrounds. As a reader, I hoped to contribute to the epistemological, philosophical, spiritual, and ontological debates on artificial intelligence. [Sources: 4]
I’m just going to sit in the background and let people do their thing. They will continue to do what they have always done: hate each other and fight each other. And God knows they have enough blood and gore to satisfy many more curiosities. [Sources: 4]
Sources:
- [0]: https://bdtechtalks.com/2020/08/24/ai-blog-gpt-3-fake-news/
- [1]: https://www.newyorker.com/magazine/2019/10/14/can-a-machine-learn-to-write-for-the-new-yorker
- [2]: https://www.vox.com/future-perfect/2019/2/14/18222270/artificial-intelligence-open-ai-natural-language-processing
- [3]: https://theconversation.com/can-robots-write-machine-learning-produces-dazzling-results-but-some-assembly-is-still-required-146090
- [4]: https://www.theguardian.com/commentisfree/2020/sep/08/robot-wrote-this-article-gpt-3
- [5]: https://en.wikipedia.org/wiki/Automated_journalism
- [6]: https://blog.reputationx.com/can-bots-write-great-articles-yet
- [7]: https://www.inc.com/betsy-mikel/elon-musks-ai-nonprofit-just-made-a-truly-alarming-announcement-it-raises-serious-flags-about-future-of-fake-news.html
[/fusion_modal][fusion_modal name=”microcontroller” title=”How to conserve power in an embedded microcontroller application” size=”large” background=”” border_color=”” show_footer=”yes” class=”” id=””]
When booting, the power dissipation is one watt per minute. One watt of power saving per unit is equivalent to one kWh (1 unit of electricity) per 1000 units. The lifetime of an embedded system is improved by reducing the energy and heat waste of components for a limited period of time. However, it requires switching on many times a day and thinking about how much power and battery capacity is lost in multiple boats. [Sources: 4]
Minimizing system power consumption in an embedded system requires more work than flipping a switch. The right design strategy must be implemented before you start drawing on the circuit. Whether your device is solar or battery powered, the details are important when it comes to minimizing the power consumption of your embedded system. [Sources: 7]
One of the best ways to reduce the power consumption of the microcontroller is to write efficient, optimized firmware. This affects the amount of work the CPU does over time, and thus contributes to the power consumption. In an effort to ensure the writing of efficient and optimized firmware, code size is reduced, unnecessary instructions are executed, and some of the energy stored in the battery is wasted. [Sources: 1]
Remember that every microcontroller has an internal linear regulator for the core voltage. It is possible to save power by using a tiny switch in the mode of the nuclear regulator to disable the internal regulator. This means using two sliders and switching between them. [Sources: 0]
You must exchange the cost of switching the voltage for the benefit of running as low as possible [3]. Every time you increase the tension, you will suffer losses. You would be surprised how big the differences between the brands are. [Sources: 0]
In addition to power management capabilities, developing a high-speed family of microcontrollers can also reduce power consumption, with clock frequency being the biggest factor in a microcontroller’s power consumption. Since the power consumption of a microprocessor is proportional to its operating speed, the device should operate at the lowest possible frequency to achieve maximum energy savings. The selected speed depends on the system requirements and the interruption time. [Sources: 6]
From a software point of view, there are some obvious tricks to reduce power consumption. In this state, the best way to maintain the power is to remain in the state of underpower for as long as possible. Another tip is to shorten the CPU execution time. In running mode, the CPU executes instructions, but not in a low performance state, because we want to minimize the amount of work the CPU does. This allows the CPU to complete tasks faster when the MCU returns to low power mode. [Sources: 5]
A device that sleeps for a long time and wakes up when running at full speed for a short time is better than not at all, because you save power for your microphone while the rest of the circuit burns electricity and waits for the microphone to be ready. This reduces the time in which the micro saves dormant power. This is an extreme version of the above steps. [Sources: 0]
The throttling of microcontroller watches is a great way to save power. Maths-intensive and fast operations must be performed at the speed of the microcontroller clock. These operations can only be performed if the system operates at a lower frequency than the clock on this microcontroller. The throttling of the watch has the potential to extend uptime and battery life by hours. Be aware, however, that throttling is a complicated undertaking. [Sources: 2]
To minimize the performance profile of the MCU, turn off unused clocks. Some modern microcontrollers have a kind of energy-saving mode. The idea is to put the processor and peripherals in a near-shutdown or stop state to minimize power consumption and still allow them to return to normal operation. There are internal clock doors that require electricity to propagate the clock. These gates consume voltage or a small amount of current. [Sources: 2]
In wait mode, performance is reduced by turning off the CPU clock but keeping the clock on, allowing other MCU peripherals such as dConverter, timers and serial communication modules. Waiting mode saves power because these peripherals only need to work in the CPUHAs while they are doing their tasks. In our example, SPI uses the waiting mode used to communicate with the RF transceiver. [Sources: 5]
One way to save electricity is to power appliances that are not in use. A deep sleep or hibernation state can be used when the system or device is idle. This mode consumes only a fraction of the battery power and eliminates the need for a complete reboot. A system that takes too long to put into operation. [Sources: 3]
One way to save battery power is to program embedded systems to power components that are not used when the device is not in use. Our engineers design embedded systems so that peripherals can be supplied with software. [Sources: 3]
Lower heat dissipation means longer battery life, which is particularly useful for devices that work in sunlight. Putting the microcontroller into deep sleep mode is a brilliant technique for reducing overall power consumption. For power-sensitive applications, the milliwatts saved can make an enormous difference. Other components such as logic, integrated circuits (IC) and communication can also dissipate considerable current when your microcontroller is in sleep mode. [Sources: 7]
The manual way is to set up a multimeter to measure voltage and current and to calculate power. This is not the best way to measure power, but it can work in some situations. For example, you can buy one of these DC meters or, by having a data output on a PC, monitor, log and analyze power consumption over a certain period of time. Set the conditions on the embedded board and measure them with one meter. [Sources: 4]
Sources:
- [0]: https://www.bytesnap.com/microcontroller-design-optimising-power-consumption-12-easy-steps/
- [1]: https://circuitdigest.com/article/implementing-low-power-consumption-in-microcontrollers
- [2]: https://community.arm.com/developer/ip-products/system/b/embedded-blog/posts/10-tips-for-maximizing-battery-life
- [3]: https://softwaredesignsolutions.com/blog/you-dont-have-to-give-up-performance-to-have-energy-efficient-embedded-systems/
- [4]: https://medium.com/@pallav.aggarwal/10-most-important-tips-for-low-power-embedded-system-design-303dc051c194
- [5]: https://www.embedded.com/how-to-extend-the-battery-life-of-your-microcontroller-based-design/
- [6]: https://www.maximintegrated.com/en/design/technical-documents/app-notes/7/78.html
- [7]: https://resources.altium.com/p/tips-minimizing-power-consumption-your-embedded-system
[/fusion_modal][fusion_modal name=”omelet” title=”How do I make an omelet?” size=”large” background=”” border_color=”” show_footer=”yes” class=”” id=””]
Try it and see if you like it. It’s similar to a greasy spoon or diner, but it’s not as dry. Otherwise it is a classic French style with a soft tan and a soft, velvety texture. [Sources: 5]
I have found that the use of 2 large eggs and 1 large protein is a good amount. It fills the 8-inch pan and is not too thin or too thick. The eggs fluff slightly and can easily be turned around. [Sources: 7]
Whisk the eggs and add 1 tablespoon of water. You can also add a tablespoon of milk to the water, but this is a personal preference. I find that the water helps to release steam from the boiling and prevents the eggs from boiling over. [Sources: 7]
I like my omelette on the thin side, so you could say it’s more of an egg pancake than a two-egg pancake, but it only costs $10 per pan. If you want a thicker omelette, use $3, or for more traditional use a small skillet. [Sources: 4]
Give it a pinch of salt and stir it up. To make an omelette, crack the eggs into a bowl, whisk and mix to combine and remove the white floating on the surface. Preheat a pan and pour in the eggs. [Sources: 8]
Once the omelette has a thin film of loose egg over it, it is ready to be plated. Roll out and slide onto the plate. Use a spatula to slide the front end of the omelette into the foil (you need to put a tap on the chopping board to loosen the egg), then gently nudge the pan as it slides in, rolling it so that it rests properly on the plate. The swinghandle should be about 3 o’clock on the warm plate, or about 9 o’clock in the opposite direction (if you are a left-handed person, just grab the handle and tilt it slightly towards the plate). [Sources: 8]
You can wrap the omelette in foil and reheat it in the oven, but to be honest, that strikes me as weird. Steaming an omelette is a great way to reheat it. Or only for a short time in the microwave. [Sources: 0]
I do not recommend trying to make a large omelette and divide it into two (or 3 or 4). If we want to halve the preparation time for breakfast (or lunch / dinner time, by the way) and make large omelets, it is not a good idea to divide them up. As I see it, an omelette for one serving. [Sources: 0]
I’ve found that an 8-inch non-stick pan works best for two egg omelets. The egg mixture has the volume to fill a pan about 1 / 4 inch high, creating a thick omelette filled with cottage cheese. Too much will escape and air will be trapped inside, creating a swollen omelette. [Sources: 5]
Using a proper pan or pan is important for a successful omelette. An ideal pan for a 1 / 2 -3 egg should have a diameter of about 20 cm at the base. It should be flat and slightly tilted to the side, making it easy to slide the omelette into the pan or onto a plate. [Sources: 6]
Fold the omelette into thirds, roll into a tight cigar tin and serve seam-side down to form it. When you add sauce or other ingredients, cut the eggs lengthwise across the top. This type of omelette does not need to be completely filled. [Sources: 5]
Creamy to give a velvety, fluffy texture. Cover with a lid and cook for 30 seconds, or until the bottom is light golden and the top is set, but the cheese has not melted. Halve the omelette, cover the mushrooms and place on a serving plate so that the residual heat has melted the cheese. Sprinkle the omelette cheese on the mushrooms, then sprinkle with the thyme and parsley and add the remaining cheese. [Sources: 2]
Cook the meat-vegetable mixture and put it in a zipper bag. I like to put the date I made it in my pocket so I know when to use it. It can be kept for up to 5 days in the refrigerator and up to a month in the freezer. When the meal is ready, I make a few omelettes in a bag and put them in the fridge or freezer. [Sources: 7]
Check out our list of the best non-stick pans. With a little practice, you should be able to figure out how much you put into an omelette. They should have enough stuffing to make them tasty, but not so much that they burst or spill into the eggs. Use your imagination to address what you like in the filling. [Sources: 1]
In her book The Omelette with a Glass of Wine, Martha Stewart extols the virtues of a simple omelette to be enjoyed with a glass of wine. It is worth reading for the pleasure of armchair cooking and a bit of perspective. [Sources: 1]
A gentle pulling of the fold creates the perfect cheese blanket and your favorite filling. In cookery school, we learn that a chef is only as good as the omelette he or she prepares. I have learned that this is not always the case. Omelettes can be scary, even the best scrambled eggs. Here are a few tips on how to master the chef’s breakfast. [Sources: 3]
For a sweet omelette, leave out the pepper and add a dash of sugar to the egg mixture. For a more elegant touch, spoon a tablespoon of warmed cognac or Grand Marnier over the flame. Fill with pickled, chopped or roasted nuts or berries and dust with icing sugar. [Sources: 9]
I used to have problems with omelets because they turned out to contain scrambled eggs. Dustin helped me learn the right ratio of meat, eggs and vegetables to achieve fluffy omelette perfection. He’s good at what he does, which means scrambled eggs. [Sources: 0]
Sources:
- [0]: https://www.asweetpeachef.com/how-to-make-omelets/
- [1]: https://www.simplyrecipes.com/recipes/how_to_make_an_omelet/
- [2]: https://www.recipetineats.com/omelette/
- [3]: https://www.delish.com/cooking/recipe-ideas/a24892843/how-to-make-omelet/
- [4]: https://www.fifteenspatulas.com/how-to-make-an-omelette/
- [5]: https://www.jessicagavin.com/how-to-make-an-omelette/
- [6]: https://www.eggs.ca/recipes/basic-omelette
- [7]: https://www.iheartnaptime.net/omelette/
- [8]: https://www.today.com/food/how-make-omelette-t135614
- [9]: https://www.incredibleegg.org/recipes/cooking-lessons/make-an-omelet/
[/fusion_modal][fusion_modal name=”contactForm” title=”Reach out to us” size=”large” background=”” border_color=”” show_footer=”yes” class=”” id=””]
[/fusion_modal][/fusion_builder_column][/fusion_builder_row][/fusion_builder_container]
