MACHINE LEARNING goes physical [2021 trends]

We all got used to (not necessarily appreciative of) machine learning recommending products, sorting our social media feeds or matching us with potential dating partners. However, the common factor of all this AI use-cases has been that it tries to influence our lives through software interface. Experts believe 2021 will change this, with machine learning entering also our physical lives. Not convinced? Read on.

Most of the autonomous machines have been developed for close environments of the factories, warehouses or other sites that are insulated from real human humming. The reason why machines were not allowed “out among humans” is that constrained environments set the sensing and intelligence need bar just low enough for machines to actually exceed it. However, recent development in reinforcement learning, NLP, computer vision and auto-ML areas have opened room for constructing intelligent machines that manage to operate also in open society.

The main point of change is that above mentioned cocktail of AI approaches gives machines chance to learn beyond their original scope of responsibilities (may the need be and opportunity be granted by its creators). As result, we will meet robots moving along to fulfill their duties. The interesting part of this change is that it all will happen without the consent of individuals passing by. (Can you object to street sweeping machine, it if does not hurt anybody and does not destroy anything of the value?) One should also admit that highest goal of first ML moving objects “out in wild”, most likely, will be to avoid humans first place. So, machine learning will not only get physical, but also (somewhat) ignorant.

 

 

 

QUANTUM gets networked [2021 trends]

Quantum computing (QC) is not a new-kid-on-the-block. If you still seek easy way to wrap you head around quantum computer, here is short take on it from one of my previous posts. However, for QC to cross the line from cool to useful, it still needs to accumulate more QuBITs (basic operating capacity units) to arrive at (range of) thousands; where even the best quantum computers are still below 100 of them for now. Until they get there, quantum computer remains fancy, too costly brag-item of the lab or company annual report.

On the positive note, human kind seems to take lessons from ordinary computer’s history. The real burst in (commercial) heavy computing launched when companies had stopped trying to build their own gym-size server rooms and honed the power of networked, cloud orchestra of “ordinary” PCs. Same way out of the trouble, Quantum computing will in 2021 enter era of networked clusters. Because, as complicated it is to amass large Qubit units in single system, our internet infrastructure makes it possible (if not down right easy) to gear distant smaller QC’s into powerful assemblies.

This “let’s connect” trend will also incentivize teams already daring to build smaller Qubit instances but failing to catch-up with IBM-like behemoths race for record QuBITs in single machine. Before they had no “commercial value” behind building small QC.  Now they can rent their QC computational power in small chunks and on demand. Ok, but who cares?

Ongoing rise of QC clusters will certainly raise eyebrow of the cryptographic community. As cloud quantum computers will make it easier to crack through some recent all-mighty cypher standards. Teams striving for complex simulations will meddle their hands, too. Ultimately, networked QC’s might accelerate also quantum computing shift towards mainstream.

 

 

 

REMOTE WORK goes from bare minimum to … [2021 trends]

The cynics claim that 2020 has taught us what are 3 important life pillars: Health, Family and stable internet connection. 😊 Yes, Teleworking, home office or remote work, call it whatever you prefer, has undoubtedly become THE most important social trend of 2020. And it will certainly carry on in next years to come. As much as there are some of us, who crave for coming back to office (or often rather back to office’s coffee machine or bean bags), quite a few people feel some relief in working from home. Cutting 2+ hours of communing back-and-forth, dropping detours from work to school or Kindergarten to pick the kids up or simply having garden or park close that you can enjoy more when working from home.

Certainly, work over the distance is not 2020 invention, most of us spent some days of home office before COVID ever emerged. But more often than not, this felt in past more like garden-leave or (almost) sick day not worth officially applying for. Also some organizations were more inclined on remote collaboration of teams that other. For example Wikipedia, in its essence, is remote working. Most of it contributors has done all their work for Wikipedia from home (or other remote locations). Wikipedia turns in 2021 (unbelievable) 20 years old already (!) Hence, remote work is here for decades. So what is the going to be change in remote work in 2021?

Well, firstly the abrupt shock of ending in home-office (often literally) from night to morning, has thrown majority of the teleworkers into emergency mode. Not to much surprise PC monitors had been the 2nd most shopped-for goods in first 2 weeks of pandemic. And here comes the first important trend: As much as you can purchase additional screen (or laptop first place) to convert your room into new home-office, your software does not adopt over the night. Yes, Zoom, TeamViewer or similar tools have went through sudden spikes of demand, but what I mean here are the office platforms as well. Most of the banking teller applications don’t run over the cloud, and (for security reasons) you need to be logged into bank infrastructure for applications to work. Majority of the public offices did not have infrastructure or software platforms designed in a way that remote logging into would be even possible. Thus, 2021 will be year of adjustments to SW making the remote work possible (or not ridiculously cumbersome) first place.

However, the changes will not stop there. Popular collaboration tools would need to heat their R&D to enable functions in their applications that will (at least) emulate feeling of team cohesion, sense of belonging. Virtual parties, all-hands meetings and other aspects of work-related human huddles of past need to find their digital representation as well.

Some companies shoot even higher. In many corporations (including IBM as example) they have taken extra step to employ Artificial intelligence to predict compatibility of working together (even if not meeting in person). Also several initiatives proven the concept of detecting emotional state of the person from the meta-data of email communication (so not reading content of messages, but rather observing how your emailing pattern in terms of time-to-respond, length of the email and percentage replies, … evolves for individual). Though EU’s GDPR will probably raise eyebrows on such an attempt, non-EU implementations are already rolling their sleeves up.

To add something on the positive side of the AI’s role in this trend, research has been also commenced on how AI could mimic “watercooler moments” of unplanned encounters leading to ideas cross-pollination. Here the artificial intelligence predicts both the right timing as well as most likely subject to bring you “Heureka” moments into (otherwise) lonely home office session.

To sum it up, remote work in 2021 will would progress from basic (thrown-into) necessity up to something alleviating pains and pitfalls of social distance to your co-workers. Software applications (and sophisticated algorithms) will sprint to close this gap (and earn penny here and there for that). This will put even more leverage for employees desiring to work from home permanently. And as with many other work environment swings, it will crash with how much the managers are ready to let go for these new trends. Thus, if even after 2020 you are not allowed to work from home, it smells like your superiors’ phobia. <in that case don’t forget to read the 2021 recruitment trend as well>

 

 

 

 

Does your insurance cover HITTING A SATELLITE? [2021 trends]

Hey, watch your way! You nearly hit the other vehicle! No, this is not replica from driving school scene. These words will echo more and more often out of the space mission control centers. The sky is getting extremely crowded and you should correct your image of Earth-surrounding space: From vast hollow room to something more like morning commute traffic jam.

Sole Musk’s Space-X satellite program emits more than 12 000 (!) individual satellite units, not even counting all the military, weather and navigation ones. Orbits for possible launches will get scarce, the same way that FM spectrum frequencies were limited for radio stations in 1990s.

Besides the fact that mutual coordination (and interference prevention) becomes more and more complex, this satellite extravaganza will be taking its toll on other, serious aspects of science, too. Professional astronomers will have to live through vast light smog. As most of the launched satellites encompass solar panels to “fuel” the operation and its communication units. And those trying to sent something upwards, will have to count too many trajectories to smuggle through orbiting traffic.

Luckily, companies are not just waiting idle for this issue to be addressed. Some proactively mount light shades on satellites, while still others try to launch into orbits less detrimental for astrophysics observation. But low latency, global internet coverage (being recently the main urge behind the satellite launch programs) will be too much of the temptation. So watch out any time you want to send somebody to Mars. They might get seriously hurt in crash accident.

 

 

 

 

Yet ANOTHER SENSE we surrender to computers [2021 trends]

Ready to give away another human sense to computers? Well, it’s been more than 100 years since we digitalized our former senses, namely eyes (through photos and then videos) and ears (through gramophone and all following sounds systems). In the meantime, with computer vision, speech recognition and (Alexa-like) voice assistants, machines mastered these senses (almost) at par with our systems. Now the machines got already pro in sound and sight, they are ready to digitize another item of the holy 5 senses (Ok, Ladies, I correct: 6 senses). The smell will become another sense acquiring digital encoding. The pathway will , most likely, resemble previous senses digitalization. First, machines will learn to reproduce (smells), then to detect them and later to act on specific smell patterns. Don’t get mistaken, digital scent is no sci-fi anymore. Companies like Olorama or Aromyx are already developing smell simulators. Also, with vast records of online fragrance purchases, your smell preferences might already lay encoded in data vaults of Amazon and alike.

You might not realize, but the whole digital smell initiative is not a pointless, AI self-indulgence. Unlike other animals, humans severely underuse the olfactory “muscle”. So, as scents become another (software) interface, maybe we seize our final chance to close the olfactory gap to other species.

But this wont’ be just a nice, aromatic experience. As with sound and visual look, also the smell can be unique (human) identifier. Legal rule books, thus, will need to adapt to protect individual (smell) privacy as well as somebody lock-patenting basic natural smells (like rose). And your pedometer might as well be replaced one time with under-arm smell sensor. Well, I smell some action ahead …

 

 

AI will DETECT DECEASES at their earliest stages [2021 trends]

Previous year left us with (large) choice of what do you believe COVID has altered the most. Many industries have gleamed (e.g. video conferencing), some folded cards and gestured in resignation (e.g. hotels). Besides the ingenious pharma effort in vaccine development, there is less obvious 2020 hero in health sector. Artificial intelligence managed to prove it can detect diseases at their out-break, before tardy human-based detection networks can even put their protective cloth on.

You may have heard that smart watches (enabled with oximeter) could detect COVID from oxygen level in blood readings (and AI algorithm behind), but news of detecting the COVID from cough might not have reached ears of all. Yes, within few weeks after COVID outbreak there has been AI project completed that detected COVID from mere cough sound recording of the patient, with accuracy beating Antigen tests and without infected person going anywhere (which is benefit of itself).

But he lessons to take here is not anecdotal, rather fundamental. If there is something that Machine learning is strong in, then it is detecting something for which you have annotated examples of. So may it be another pandemic, or just important enough illness, if it has visual or audial symptoms, it can be immediately geared into AI detection. Now think of all the (PCR test cases) waste and human effort that could have been saved with COVID tests if all you had to do was cough into microphone.

Yes, artificial intelligence has proven that it is ready to take over significant burden of mass (symptoms) testing. And it can claim early stage detection role in health-care systems. If only doctors let go …