​As we have noted previously, AI systems are not just rows of processor cards with blinking LEDs but are systems that must be almost spoon-fed information upon which algorithms provided to them, make decisions when queried.  Before such systems become viable, they need to be taught, using vast amounts of data and ‘teaching’ algorithms that are provided by data engineers and are separate from processing algorithms.  That said, before the systems are taught, the data must be ‘organized’ into data sets that the system can recognize, which must be accomplished by humans sitting in front of computers.  Each of these processes are different, involving both humans and computers and computers alone, but they all have one thing in common and that is that they all use energy.
While the current AI focus has put such systems in the public eye, but computers have been sucking down power for years, and while models can make estimates of what percentage of the global power grid they use and how much it increases each year, there is no specific monitoring, or is there a way to come up with an accurate estimate.  The models estimated that in 2018 ~1% to 2% of the global power supply was used by computers, with 2020 as high as 4% to 6%, and models predict that at that rate it will reach anywhere from 8% to 20% by 20230, and most of that modeling was done before the recent AI fanfare and funding.
Carbon emissions are certainly a concern as demand increases but there is also a longer-term concern that with that kind of consumption growth, capacity will not meet demand.  AI systems will get more efficient and consumption/cpu will decline, but if CPU volume growth continues or accelerates, as it seems to be current doing, who will get ‘power priority’, the average TV viewer or the AI system that is running the local telecommunication system?  Of course, this is a bit sensational, although still a concern when trying to balance a green environment and continued demand for energy, but Microsoft (MSFT) seems to be thinking ahead a has signed a deal to ‘guarantee’ its potential energy needs in the future.
According to reports, Helion (pvt) a start-up funded by Sam Altman, the current CEO of OpenAI (pvt), the company that developed ChatGPT, Dustin Moskowitz, a founding member of Meta (FB), and Reid Hoffman, the Co-founder of Linkedin (MSFT), has agreed to provide Microsoft with power it generates by 2028.  While this might not seem unusual, Microsoft likely has an army of negotiators working with utilities across the globe, this source is a bit different as Helion’s technology is based on nuclear fusion.  Helion is so positive that they can accomplish this goal that they have agreed to generate at least 50 Megawatts in the following year or pay a penalty, all of this with the understanding that to date fusion technology has not produced any electricity.    Helion has even signed an agreement with Constellation Energy (CEG), the owner of the largest nuclear power plant in the US, to manage the Helion project and market its output.
Helion has promised to demonstrate a working prototype next year, with Mr. Altman adding, “Our goal is to make the coolest tech demo in the world.  Our goal is to power the world and do it extremely cheaply.”  Of course such optimism comes after one pumps $375m into Helion but he insists Helion will be able to deliver power to the global grid by 2028 and checks in on the company roughly once a month.  The fusion industry did receive a boost last December when scientists at Lawrence Livermore National Laboratory were able to generate more energy that they put in during a fusion reaction, but on a net basis that is a far cry from being a commercial fusion producer at a cost that is competitive with other energy modalities.  Is Microsoft drinking Altman’s very expensive Kool-Aid after signing a multi-billion-dollar partnership with Altman’s OpenAI, or do they know something no one else does?  We should know by the end of next year.

As artificial intelligence continues to appear in the headlines, both positive and negative, there is one group that is certainly benefiting from the publicity.  Companies, Associations, and universities are hiring lobbyists to make their views known to the federal government.  Over 120 public and private entities lobbied the government in the 1st quarter of this year, with 158 doing the same for all of last year and 30 in 2017.  The more obvious companies that have a major stake in the technology, such as Amazon (AMZN), IBM (IBM), Microsoft (MSFT), and Nvidia (NVDA), were joined by the American Bar Association, the Consumer Technology Association, the Recording Industry Association, and the Screen Actors Guild/AFTRA, along with Carnegie Mellon University, Case Western Reserve University, Harvard University, Southern Methodist University, and Washington State University, in making their opinions known to the government, and, of course, a number of pure AI companies themselves.
 
A number of large companies suddenly realized that they were behind in the ‘AI race’ and have announced plans and projects to find ways to integrate cognitive AI into their platforms, with Microsoft leading the way with plans to integrate ChatGPT into Bing search.  A number of these companies cut Ai project staffing at the end of last year as the technology space  contracted, especially those Ai projects relating to Ai ethics, are now planning to build Ai project staffing to maintain an active level of competition with their peers., while the US Chamber of Commerce, the largest lobbying organization in the US, spent ~$19m in 1Q, including establishing a taskforce on AI in the House Committee on Financial Services, implementing the National Artificial Intelligence Act of 2020 (primarily ‘studies’ and standards research), and began drafting bills relating to automated vehicles.  At least three major insurance companies lobbied for congressional efforts to better understand the commercial use of AI, and a number of universities lobbied to support the Army AI Center and distributed AI applications for defense, as they likely get funded as part of the center’s project budget.
 
​

​As we noted previously, there has been a a bit of backlash from a large group of researchers and academics concerning the rapid pace at which AI systems are being developed, with an emphasis on the ethical use of such software, with some of the more extreme groups pushing for an investigation into OpenAI, the developer of ChatGPT, citing GPT-4 as ‘biased, deceptive, and a risk to privacy and public safety”, while lobbying for ‘algorithmic transparency’, as comments by IBM that they would be replacing 7,800 workers with Ai systems over the next few years sparked news organizations to create headlines like: “AI could replace equivalent of 300 million jobs – report” (BBC – 3/28/23), ChatGPT: the 10 jobs Most at Risk of Being Replaced by AI (business Insider- 4/9/23), or “How AI is Replacing Jobs 2023: A Look at the Statistics (Gitnux Blog – 3/23/23) that cited “AI is estimated to replace 85m jobs by 2025, but 97m new jobs will be created due to AI”, sort of a bad news/good news headline that would likely incite almost any labor-related organization, and send publicity seeking Congress men and women into a media frenzy.
It is obviously quite important that the government understands the Ai industry and the impact it can and will have on society and US economics, but we fear it will become both the media calling card and a path to screen time for politicians, and then fade away when a new hot button issue takes its place, just like the Metaverse did last year.  AI, particularly the ethics and legalities of AI are extremely complex issues and the odds are that politicians are not going to spend a lot of time on them unless it affects them directly, perhaps when it affects a large campaign donor.  Lobbyists are trying to push their client’s agendas to politicians now that the public has been made aware that they now have another ‘fear’ to consider, without the positive balance that shows what AI is able to do for society.  But eye-catching headlines are not usually made from “AI Could Help to Cure Cancer” over “AI Will Take Your Job Soon”.  At best, we can expect the government to form more committees to study the problem when we could use AI to end Gerrymandering by balancing population districts, polling constituents to find out how they feel about a variety of topics, or dig through local, state, and federal budgets to look for fraud, overspending, or projects that benefit only a small portion of the electorate, the sort of things politicians are supposed to do.  Perhaps the better headline would be, “Politicians to be Replaced by AI by 2028”.

Ai is a bit far afield from our usual hunting ground, consumer electronics, but recent stories in the press about ChatGPT have sparked interest from some of our readers toward the topic, and without even stretching a little, we can easily draw the increasing connection AI has to consumer electronics.   Most consumers’ perception of AI is a large bank of servers, lights flashing, churning over billions of data points in order to arrive at a conclusion, but that is only one part of the AI story, and usually the part that has little to do with consumer electronics.  But some form of AI is built into many consumer electronics products and performs numerous functions, some of which we never see or hear, while others are commonplace.
Among the most common “AI” tools is the spell checker usually included in word processors, an outgrowth of Ralph Gorin’s 1971 SPELL program that checked letters in a word against a list of correctly spelled words by comparing single or adjacent letters, and while early spell checkers were limited to Mainframes, they migrated quickly to PCs in the early 1980’s.  These relatively simple systems were based on a comparisons, which would imply that the more words to compare against the more accurate the checker, but it was found that as the word lists grew over 100,000 the number of misspellings increased. 
The problem facing early spell checkers is that they were not able to discern context, so they faced difficulties with words that sounded the same but had different spellings and different use cases, such as ‘there’ and ‘their’.  As local computing power and storage increased, systems were designed that could evaluate context by comparing phrases or sentences against grammatical rules that were embedded in the system and could determine the correct spelling of a word based on both the word list and the rules, but data scientists found that as processing power and storage increased further, they could enter millions of data points that AI systems could ‘look at’ to better understand context, not only for language, but for almost anything, leading to the development of AI algorithms that fall into three classes.
Supervised – The most common category, where labeled data is feed to a system and is used to train the system to predict outcomes as it is given new data.  Typical systems uses decision trees that test the data using sophisticated ‘if…then’ nodes to move through the decision process until a conclusion is drawn that is based on all of the tagged data that the system has seen.  Systems like these are the basis for systems that have been in the press in recent months, such as those that can ‘paint’ a picture based on the style of a particular painter, having been fed the digital images of all of that artists work.  The system can create an image based on the user’s request using the characteristics it has recognized for that artist.  In its simplest form we asked an AI art generator to paint a picture of daisies in a vase using a generic style and Figure 1 is the result.
Unsupervised – This category does not use tagged data, so the algorithm must evaluate the relationships between the data points without specific definitions, which they tend to do by creating clusters of similar data around central data points, and using a number of methodologies to evaluated the clustered data.
Reinforcement – This category is a bit like training an animal, as the system uses an ‘agent’ to perform an action within a certain ‘environment’, and then rewards the agent when the action is completed.  The agent uses the reward (positive or negative) to help it with the next iteration of the action until the environment ends the process, just as rewarding an animal for a correct response helps the animal to understand what the correct response is.

​CE products vary widely in their use of AI systems, from virtual assistants such as Amazon’s (AMZN) Alexa  or Apple’s (AAPL) Siri to embedded Ai systems such as those in your TVs.  Virtual assistants ‘learn’ about your behavior based on the tasks they are asked to perform, so, in theory, they should be able to respond to requests with increasing accuracy as they learn an individual’s patterns.  TV AI is more technical in that it can sample incoming images for patterns and make adjustments on-the-fly to emphasize or de-emphasize a particular aspects of the image based on the user’s settings, or remove ‘noise’ or scale an image from one resolution to another.  These systems can process images on a pixel-by-pixel basis and with TVs generating ~30 frames each second, that comes to 14.9b pixels that can be processed each minute, but the AI component also ‘learns’ about each image, rather than comparing to a built-in list as in older sets, and can make better adjustments based on what it learns.
Both of the applications mentioned are common to CE products and in many cases are integral to their function, but AI continues to develop and will become more integral in CE products and daily life.  Having a background in the music industry, we were intrigued by Google’s (GOOG) MusicLM system, a bit less sophisticated than the artistic systems mentioned above, but nonetheless amazing in its ability to create music from text.  In its simplest application the words “acoustic guitar” or “flute” generate the following sound clips[1]:
 https://google-research.github.io/seanet/musiclm/examples/audio_samples/10s_samples/instruments/acoustic-guitar.wav,
 https://google-research.github.io/seanet/musiclm/examples/audio_samples/10s_samples/instruments/flute.wav
but more interesting is how it generates clips from the text “blues”, “West coast Hip-hop”, “East coast Hip-hop”, or “Reggae”
 https://google-research.github.io/seanet/musiclm/examples/audio_samples/10s_samples/genres/blues.wav
https://google-research.github.io/seanet/musiclm/examples/audio_samples/10s_samples/genres/west-coast-hip-hop.wav
https://google-research.github.io/seanet/musiclm/examples/audio_samples/10s_samples/genres/east-coast-hip-hop.wav
https://google-research.github.io/seanet/musiclm/examples/audio_samples/10s_samples/genres/reggae.wav
However things get even better when the system gets more complex text that does not specify a genre but a caption, such as “beach in the Caribbeans” or “escaping prison”,
https://google-research.github.io/seanet/musiclm/examples/audio_samples/10s_samples/places/beach-in-the-caribbeans.wav
https://google-research.github.io/seanet/musiclm/examples/audio_samples/10s_samples/places/escaping-prison.wav
with applications in TV, gaming and advertising quite obvious.  Not only is the system able to create genre related music or even contextual music, but it has also learned about diversity across all forms of music and with the same text token “progressive rock guitar solo” came up with a number of different guitar solos based on its learned data.
https://google-research.github.io/seanet/musiclm/examples/audio_samples/diversity-samples/progressive-rock-guitar-solo/same-text/0.wav
https://google-research.github.io/seanet/musiclm/examples/audio_samples/diversity-samples/progressive-rock-guitar-solo/same-text/1.wav
https://google-research.github.io/seanet/musiclm/examples/audio_samples/diversity-samples/progressive-rock-guitar-solo/same-text/2.wav
 
 
While this is only a single AI application, there are many, including the famous/infamous ChatGPT and its many clones, that will find their way into CE applications regardless of whether we believe they are useful, helpful, legal, moral, or justified as they represent ways in which content can be created without the expenses associated with human involvement.  We have given up moralizing about the potential ‘disintegration of society’ that AI has the potential to cause, since almost all generations have heard the same thing about some technological change (TV, video games, CDs, etc.), and society has found a way to adapt to the change.  However, as only a small portion of humankind has the desire to challenge the validity of textual or visual media, it will take some time before we understand how AI will affect the fabric of human creativity.  Will it turn us into a society of eloi[2], allowing AI to keep us docile with a constant supply of curated entertainment, or will it free us from the burdens of everyday life, allowing our creativity to blossom?  Honestly, we don’t have a clue but to paraphrase Bette Davis “Fasten your seatbelts, it’s going to be a bumpy ride.”


[1] These sound clips reference the Google MusicLM site links for each file.  Ctl+Click should play each individual file.

[2] “The Time Machine” – H.G. Wells 

It is Chinese New Year, which runs all this week, so to celebrate the Year of the Rabbit, we asked two AI websites to create original artwork using the phrase “Year of the Rabbit”.  Here are two examples of what was generated by these sites, both of which are free, using the same text.  With considerable new coverage appearing toward AI assisted art and writing applications that the general public is just becoming aware of, we thought it appropriate to generate our Chinese New Year art through such systems, rather than through the usual Google (GOOG) image search.  Are they better, more appropriate, or more to the point than original (human) art?  “Art is in the eye of the beholder, and everyone will have their own interpretation”[1]


[1] E.A. Bucchianeri – “Brushstrokes of a Gadfly”

Magic Leap (pvt), the Florida based AR company that has been one of the most successful AR companies in terms of raising capital, has indicated that the Saudi Arabia Public Investment Fund ($620b as of 1Q ’22) has increased its stake in the company, now holding a controlling interest.  The deal, which consisted of $150m in convertible preferred and $300m in debt, and gives the fund the right to appoint 4 of the company’s 8 directors.  Magic Leap, which was founded in 2010 by Roby Abovitz, has raised over $3.5b through 28 investors, including Google (GOOG), Alibaba (BABA), Qualcomm (QCOM), AT&T (T) and a number of VCs, but went through some difficult times in 2020.  The company was forced to lay off ~1,000 employees (~50% of staff) when sales of its Magic Leap 1 AR headset, released that year failed to materialize as expected, and rumors that the company was looking for a buyer turned out to be untrue as the company went back to private investors for more capital.  The company then changed focus (most of the layoffs came from the consumer segment) to the enterprise market, and the founder and CEO gave up the reins to a new CEO in 2020.  Since then, he has relinquished his board seat to found a new company, Sun & Thunder (pvt), that creates Ai characters (“SynthBees”), a Metaverse engine, and “some really cool and creative things”, according to the company.
Magic Leap’s initial headset problem was price, and while the device (Magic Leap 1) was quite sophisticated, it cost ~$2,300, which was hard to justify in the consumer market in 2018.  Since then the company announced the Magic Leap 2 in January of 2021, but did not release the device until September 29 of this year.  While the new device itself costs ~$3,300 (~$5,000) for the complete package), and is the most expensive AR device released in 2022, it is designed to be used across a number of industry verticals where considerable software has been developed to integrate the device with existing applications such as IT tools, 3D modeling, training, collaboration, technical guidance, and metaverse creation.  The device has considerably more internal storage capacity than similar devices, a positive for quick responses to user needs, although the display uses LCoS (Liquid Crystal on Silicon), an LCD based technology that is rarely used in such devices, having been replaced by Micro-OLED or Micro-LED display technology.
The Saudi confidence in the company is a bit surprising considering their original investment, which we believe was made in March 2018 ($461m) has seen a number of ups and downs but given the pre-money valuation of Magic Leap of between $1.5b and $2b, they seem to be optimistic about the longer-term prospects for AR and Magic Leap itself.  If the Saudi’s do take four board seats, there will have to be some reshuffling, as those seats have been held by Alibaba, Qualcomm, and Google representatives, but with over $700m invested, the Saudi fund has the right to run the board.  All in, the Saudi Fund, which remains under the control of Mohammed bin Salman, the Crown Prince and Prime Minister of Saudi Arabia, has stakes in other tech companies (Uber (UBER), Capcom (9697.JP), etc) but also has made investments in a number of US politically linked companies, particularly those of Jared Kushner, triggering issues surrounding Kushner’s ongoing support for the Saudi government, despite human rights violations.  There has been little controversy over the existing Saudi investment in Magic Leap, but given the political climate, that could change quickly given the company’s status as a US unicorn.,

The State Internet Information Office in China has released a provision relating to the use of “Deep Synthesis of Internet Information”, which more loosely translated would be the use of AI in media services of any type on the internet and in other broadcast venues.  What this refers to is the use of artificial characters that might look like actual persons or are life-like.  This has become a relatively common practice in China, where artificial avatars, who mimic existing media personalities, are used to report the news, or make public commentary.
The new rules force those using such forms of information dissemination to visibly identify them as such, and to monitor the information that is synthesized by AI-based news processors.  In many cases AI is used to sift through large volumes of news from a wide variety of sources and synthesize a short summary of a story that seems to be of consequence.  This is done through the AI system, which looks for recognizable patterns in news flow, and puts together a composite of those stories that it or others have selected by combining source information.
The Chinese government is attempting to move the responsibility for accurate reporting to the media source, making sure that the media owner knows he or she is responsible for making sure the information disseminated is accurate and legal, in that it does not violate existing laws that “endanger national security and interests, damage national image, infringe on social public interests, and disrupt the economy”.  While this seems to be an attempt to ‘keep things real’ by making sure AI-based information is disclosed as such and is legitimate, it also makes clear that the responsibility for meeting all existing laws, some of which are obviously censorship related, are followed, and that blaming misinformation or illegal (meaning different from state rhetoric) information on the AI is not a valid excuse.  Owners of the distribution source must keep concise records of any issues relating to information sources and log and report any discoveries of incorrect or illegal material generated by the AI system, but the rules state that the responsibility to meet applicable laws are the responsibility of the owner, no matter what the AI might broadcast, and that the public must know when Ai generated information is being used.
“For the public good” shows up a number of times in the new regulations, but the rules also seem to have a repressive tone, seemingly extending the government’s ability to suspend, terminate, or prosecute the accounts of those who might find themselves at odds with the state.  Given the recent protests in China, it seems that this is another way to plug up any holes in the media censorship that exists in China, and while it has technology related overtones, it is really a way of reminding the media that it lives under strict rules and technological mistakes or omissions will result in the same repressive results that would occur in any other form of media that exists on the Mainland.
Examples of AI Generated ‘Digital Twins’ are shown below courtesy of Deep Brain AI (pvt), a South Korean firm that is considered the preeminent provider of such technology.
https://renew.deepbrainai.io/renewal_resources/videos/Use+cases+%26+Industries/Media_4_CCTV.mp4
https://renew.deepbrainai.io/renewal_resources/videos/Use+cases+%26+Industries/Media_2_LG.mp4
​

WIPO (The World Intellectual Property Organization) is a body that represents 77 member nations, including China, the US, and Russia, that creates IP policy under which its members are supposed to operate and maintains an IP database for many of the members that do not have a substantial IP system.  While policy still varies from country to country, there is a topic that has become a talking point for the IP world and that is artificial intelligence.  “The Next Rembrandt”, a portrait created by an AI system that sampled 168,263 Rembrandt painting fragments coupled with a facial recognition system or Google’s GOOG) Deep Mind AI system (WaveNet) that has been used to create short snippets of “Chopin-like” music based on millions of Chopin samples, are bringing up questions as to ownership of such IP, and that creates considerable controversy, but it goes much further.

​Things get really confusing when the AI itself is the applicant for its own ‘created’ works and while discussions at WIPO and other IP venues continue, it has been up to the courts in each country to rule on whether such applications are valid.  AI developer Steven Taylor submitted an application to 16 countries last year with the inventor’s name “DABUS”.  He had no knowledge of the devices described in the application and claimed that DABUS had created two inventions after developing the general knowledge to devise such items.  The Korean Intellectual Property Office asked Mr. Taylor to amend the application using a natural person as the inventor, however Mr. Taylor did not make such a change, and the application was rejected.
The basis for the rejection was that patent law and precedents state that only ‘natural persons’ may be inventors, with a number of countries, including the US, Germany and the UK using this as a guideline for applications, although a lower court in Australia recognized AI as an inventor last year, although it was later overturned  by a higher court.  A conference held late last year  that represented seven patent offices (including the US, China, and Europe) came to the conclusion that those countries have not yet reached the level of technological sophistication that would allow an AI system to ‘invent’ without human intervention and therefore could not consider an AI as a natural person/inventor, but as AI systems become more refined and make an increasing number of decisions on their own, the definition of natural person does not seem to meet the necessary level of breadth that is needed in such cases.
In most instances where such decisions have to be made, the application is either denied or is attributed to the developer of the software that is the basis for the AI creation[1], but copywriting of works generated by AI does not seem to have been specifically prohibited, although there are many countries, including the US, that specify a copyright must be created by a human being, or as US case law[2] stated it, “…the fruits of intellectual labor that are founded in the creative powers of the mind”.  Hong Kong goes in the other direction, along with India, Ireland and New Zealand, and the UK, based on the UK’s Copyright, Designs, and Patents Act, which states: In the case of a literary, dramatic, musical or artistic work which is computer-generated, the author shall be taken to be the person by whom the arrangements necessary for the creation of the work are undertaken”.
But referring the patent back to the software ‘s author would be the same as assigning a patent to the actual writer of the patent, not the one who came up with the invention, and call into question an almost infinite number of semiconductor patents that were designed on software developed by a relatively small number of companies.  Should all of those semiconductors be the property of the companies that developed the design software?  It’s not a question we can answer and one that will get harder to answer as AI systems become more able to make basic decisions on their own. The underlying program that runs the basic functions of the AI are the framework under which the system runs but that framework allows an increasing amount of decisions to be made by the system itself as it learns.  There will be a point at which AI systems can get close to synthesizing human thought or creativity, at which point the question becomes far more difficult to answer, but there will always be the case that the software is the basis for the higher level AI decisions or creativity, which would give credence to the ‘human only’ ideology.


[1] Guadamuz, Andreas. “Artificial Intelligence and Copyright.” WIPO MAgazine, May 2017.

[2] Feist Publications v Rural Telephone Service Company, Inc. 499 U.S. 340 (1991)) 

In November 2018 the Chinese State run news agency, Xinhua (state), unveiled it 1st AI generated news anchor based on an actual Chinese newscaster named Zhang Zhow.  The artificially generated newscaster is able to mimic the actual broadcaster’s movements and speech, allowing him to be available to the news desk on a 24 hour basis, with a female AI broadcaster developed about a year later, based on another well-known Xinhua broadcaster.  These clone reporters are able to mimic complex facial and body movements that are derived from the AI being trained on the specific person’s video images and fill in for the broadcaster when an event happens quickly or the actual person is unavailable or on assignment.

In April 2016 a Japanese Advertising agency pitted a human creative director against an AI system to compete for program ideas for Chlorettes’s Mint Gum, with the AI ‘creative director’ developed by the Japanese arm of McCann (IPG).  While the human CD won the contest, although the vote was close (54% to 46%) and similar contests for two other advertising programs went to the AI.  Taking the AI concept a bit further in the advertising world are a number of Ai companies that specialized in the creation of digital models that can be used for advertising without the time and expense consumer photoshoots that are typically used.  The models shown below were generated by Rosebud.ai (pvt), a company backed by a ‘Who’s Who’ of Silicon Valley cognoscenti, and now seemingly focused on helping users generate NFT’s, with companies like DataGrid (pvt) https://youtu.be/8siezzLXbNo and NEON (pvt) https://youtu.be/2UlBFiL6noU creating even more life-like models that are more expressive than those mentioned earlier.

In 2017 Japanese advertising agency Dentsu (4324.JP) introduced AICO, short for AI Copywriter, a system that is able to generate advertising copy that could generate vast amounts of copy after researching (being trained on) specific topics and given a bit of context on the target idea.  Since then many AI based copywriting tools have been developed, a number of which are easily accessible on line.  Here’s a demo from Anyword (pvt) on “Coffee-Making Software” generated by their AI system:
“Everybody needs coffee.  Our state-of-the-art machine does not discriminate.
No matter the type of coffee you enjoy, there is a setting for it on our coffee maker.  We priced ourselves on being able to offer options for everyone, making our machine versatile enough to be used by anyone.
The process couldn’t be easier!  Choose your favorite roast from our custom selection, then choose which brew method fits your lifestyle best.  From there the AI will take care of the rest, using precision technology that won’t burn your coffee beans or brew too fast so your grounds can properly steep to release their flavor before being round.
Give us a try today!”
The ultimate (for the time being) escalation of the use of AI in the advertising world is Rozy, an incarnation of Korea’s Sidus Studio X (035420.KS), who is a ‘virtual influencer and model’ who earned close to $1m last year.  While the 22 year old influencer looks and acts like more typical influencers or pop stars that sign deals to hawk various items to those who are predisposed to follow whatever trend is current, she is not a real person and is therefore available 24 hours/day  and does not need to be flown (along with a retinue of make-up artists, wardrobe specialists, hair designers, and aides) to exotic locations for a few hour shoot, with only the background footage being necessary, which can be shot by local videographers.  She does not age, nor does she get negative publicity after heavy partying with other celebs, and she has picked up endorsements from over 100 sponsorships since her ‘creation’ in 2020, but the ultimate in AI based ‘advertising’ is expected to be MAVE, a virtual only all girl group devised by Kakao Entertainment (103260.KS) that will follow in the footsteps of ‘aespa’, the SM Entertainment (SMCE) actual girl group that has a virtual equivalent.

​All in, the use of AI in broadcast and advertising continues to increase and as the technology improves it will get increasingly more difficult for consumers to know whether they are being served by a human or a virtual copywriter, newscaster, product influencer, or even a friend, as some of the AI imaging products can be used to ‘prank your friends’ by creating virtual images of on-line videos and changing dialogue or actions.  While advertisers extol the use of AI to ‘personalize’ the advertising experience by having the systems write ‘better’ and higher volumes of copy and targeting that copy to the collected data for individuals, and news networks justify their Ai use as a cost-saving measure, we wonder how much humanity gets lost when an AI does the interpretation of an image or a headline and whether the Ai has even the smallest bit of conscience when it comes to advertising (not to say more advertisers have much).  In 1937 radio journalist Herb Morrison’s onsite broadcast about the crash of the airship Hindenburg, which killed 35 people in a fiery explosion might have been this:
“The airship is combusting and falling down with flames and smoke as it nears the mooring mast.  There are passengers screaming and the air is filled with acrid smoke”
Instead of the historic rendition that expressed a bit more emotion:
“It’s fire and it crashing! . . . This is the worst of the worst catastrophes in the world! Oh, it’s crashing . . . oh, four or five hundred feet into the sky, and it’s a terrific crash, ladies and gentlemen. There’s smoke, and there’s flames, now, and the frame is crashing to the ground, not quite to the mooring mast. Oh, the humanity, and all the passengers screaming around here!
. . . I can’t talk, ladies and gentlemen. Honest, it’s just laying there, a mass of smoking wreckage, and everybody can hardly breathe and talk . . . Honest, I can hardly breathe. I’m going to step inside where I cannot see it. . . .”

​There is a war going on that rarely makes it into the press as the battlefield is not on the ground, in the air, or in space, but inside the guts of massive processing nodes that are used to understand the nuance of language.  The algorithms that direct these processor are based on a subset of Ai that deals directly with syntax, expression, and a host of other language variables that make understanding other languages a challenge for humans and a monumental task for digital entities.  The opponents here are companies rather than political entities, and well-known ones at that, with Microsoft (MSFT), Google (GOOG) and Meta (FB) all pitted against each other with the focus of being the dominant force in the digital translation market.
There are no casualties or battle lines in this war, as it is fought with processing metrics and advertising, and that makes it hard to know who is winning, but the participants seem to have taken on a particular metric to make the public aware of who might be in the lead.  That metric is the number of languages a system can translate, which seems to be used a s a gauge toward whose system is the most advanced.  While the number of languages that a system can translate is certainly important, especially to those languages that might be considered secondary, but by far the most important metric for translation services is accuracy. 
Recently Meta indicated that its NLLB-200 Ai model has increased the number of languages it can translate to 200, which it accomplished in two years, while Google Translate is only able to work with 133 languages and Microsoft’s system translates only 111 languages, although that includes two Klingon dialects, and is clearly staking a claim as the world’s most advanced translation tool.  While the number of languages a system is able to detect is certainly easily understood by the general public, the quality of the translation, which is based on the algorithm and the sample base, is far more important and there are two ways in which that can be evaluated, by humans or by machines.  Using humans to evaluate translation quality throws subjectivity into the mix, while automated machine evaluation does not, but a machine evaluates a translation by averaging words and sentences against a human evaluation, with the idea that the closer the machine score is to a human score, the better the translation is.
With all of this being beyond the scope or desire of the general population, translation giants will continue to use the simplest metrics to give credence to their systems, but will have little correlation to real world results, as the ability of the AI to understand nuance and what to do with that understanding is really the key.  All three companies mentioned here have access to vast stores of speech, which certainly goes toward the ability of an AI to learn, but the algorithm is the key and that is something that not only grows with more resources but must change as humans better learn how to convert those subjective views into language that a machine can understand.  So the question is, does the AI need more language to read or does it need more human evaluations in order to take the subjectivity out of their evaluation?  The only way to know is…
E ninau i ka Mīkini - Hawaiian
Spurðu vélina - Icelandic
Faighnich dhan Inneal – Scottish Gaelic
quaeritur machina – Latin
Ask the Machine - English
 
 
 
 
 
 
 

At the end of last year LG (003550.KS), parent of LG Electronics (005930.KS), LG Display (LPL), LG Chem (051910.KS), and a host of other subsidiaries, announced their “ultra-large” AI model (Exaone) that is said to have the ability to capture 300 billion variables, generating a higher level of language skills than other Ai systems.  As most human knowledge is contained in language, the current thinking is that developing more comprehensive, more efficient, and faster AI systems based on language will lead to a system that will come closer to the human characteristics that seem to be the goal of AI developers.  According to LG, Exaone, rather than searching through images based on the text surrounding them, this system creates new images based on the data it has examined, and can show and describe its creations in both Korean and English.
While perhaps not the application that many AI experts were counting on, LG has used the Exaone recently to answer a question, “What would it look like if there were flowers on Venus?”  While we are sure that is not the only question asked of this Ai system, which cost many millions in R&D and training costs, it seems a bit strange that this is the direction taken for Exaone, at least at this point, but Exaone not only came up with an answer (actually 3,000 answers) but has created ‘Tilda’, a virtual “environmentally-conscious fashion designer” to help us humans to understand what Exaone came up with. 
Tilda has collaborated with Korean fashion designer Park Youn-hee to create a Fall/Winter ready-to-wear clothing collection (200 outfits) “Greedilous by Tilda – Flowers on Venus”, that is being highlighted at the New York Fashion Week running through Wednesday.  Tilda supplied the ideas about patterns based on the question mentioned above and Park took those ideas and created the collection in 45 days, a process that typically takes many months.  Tilda sourced her ideas from the 250 million high resolution images and 600 billion textual data that was feed to the Exaone.  After fashion week Tilda is planning to launch an independent eco-friendly fashion brand to deliver her message about the environment through fashion.
Officials from the LG AI Research Institute state that “Within a year, you will be able to meet Tilda’s unique fashion products and art works that embody the philosophy of Tilda both online and offline”, and the institute plans on creating more expert AI ‘humans’ (their words) that help and cooperate with humans in various fields such as manufacturing, research, service, education, and finance.  We are also planning to venture into the Metaverse, where we can communicate with the Gen Z and take part in more creative processes.”
John McCarthy, a professor emeritus at Stanford, coined the term “Artificial Intelligence” in 1955 in a proposal for a 2 month, 10 person summer research conference, according to the college, and went on to invent LISP, an early programming language that is still used for AI today, and founded the MIT Artificial Intelligence Project and the Stanford Artificial Intelligence Lab.  While we don’t know for sure, we doubt John was thinking about how artificial intelligence could be used to create a stir at Fashion Week 2022, but we expect he would have liked to spend time speaking with Tilda once she stopped talking about how drip the new collection was and how jaboni the other designers were.