Reply to this post
Both need to change to fare well.
- Baidu reported good Q4 17 results as it has put the tribulations of the 18 months squarely behind it and is now focused on becoming the preeminent supplier of AI in China.
- Q4 17 revenues / EPS were RMB23.6bn ($3.74bn) / RMB15 ahead of consensus estimates at RMB23.1bn / RMB13.
- With its new, more selective ad system in place, advertisers have returned to Baidu pushing mobile revenues to 76% of total revenues, an all time high.
- 2017 has been a difficult year for Baidu and its strategy is now shifting away from being a fully fledged digital ecosystem to focusing on products and services that are entirely driven by AI.
- This is why the loss making iQIYI has now filed for an IPO and I suspect that 2018 will see some movement around the ownership of its other loss-making e-commerce venture; Nuomi.
- Baidu’s main AI assets are Apollo (autonomous cars) and Duer OS (digital assistant) and in both of these, it is far and away the leader in China.
- How these will be directly monetised is less clear at this stage, but it is clear that:
- First: AI will have a major impact on the ability of ecosystems to differentiate their digital life services over the next 10 years.
- Second: Baidu is the undisputed leader in China with both Tencent and Alibaba miles adrift despite protestations to the contrary.
- This puts Baidu in a very strong position to partner or licence to the have nots in Chinese AI (which I think is almost everyone).
- Hence, with the core business now looking to be back on an even keel, I think Baidu represents a cheap entrance to what its likely to be the biggest investment theme of the next decade.
- Uber reported headline figures for Q4 17 that showed some progress but, in my opinion, not nearly enough given its precarious position in the US market.
- Q4 17 revenues and adj-net income were $2.2bn / LOSS $1.1bn compared to $2.01bn / LOSS $1.46bn in Q3 17.
- This is good progress but given the sizes of the losses in Q3 17 and the fact that they increased meaningfully from Q2 17, I suspect that there was a lot of low hanging fruit.
- Revenue growth remains strong at 66% YoY but all of the momentum at the moment remains with Lyft which I see as being on the cusp of causing Uber real problems.
- Uber is still dominant in its home market (USA) with 66% share but this is substantially down from the 80% that it held at the beginning of 2017.
- Ride hailing businesses are marketplaces and as such are subject to the rule of thumb that I described over two years ago which still seems to be holding firm.
- This rule of thumb states that a company that relies on the network must have at least 60% market share or be at least double the size of its nearest rivals to begin really making profit (see here).
- Hence, I see 2018 as the time when Uber needs to begin looking at making some money and at the same time ensuring that Lyft bleeds badly just to stay in contention.
- Uber needs to neuter Lyft now because when it comes to autonomous driving, Lyft is a long way ahead via its relationship with Waymo.
- Should things stay the way they are, then Uber could be in real trouble once autonomous vehicles start making a real appearance in the market.
- Fortunately, this is still some way off but the threat is there and 2018 needs to be a year where Uber re-establishes its dominance in the home market especially after embarrassing loss of both China and Russia.
Reply to this post
Siri’s fragmentation could be her undoing.
- Apple’s focus on devices has meant that there are multiple different implementations of Siri which is starting to cause problems.
- If the digital assistant is to become the next generation OS, then it is Google, Baidu and Amazon that are in the driving seat as history has repeatedly shown that software fragmentation is extremely detrimental for the user experience.
- Fortunately for Apple, I think that this is a problem that it should be able to fix.
- Put simply, the problem exists because of the decision that Apple has made to run Siri locally on devices rather than have the intelligence sitting in the cloud.
- This is an understandable decision from a company so rigidly focused on devices and differentiating with security and privacy.
- However, this decision is now beginning to show some problems.
- Siri is now present on iPhone, Apple Watch, Apple TV, Mac, iPad and HomePod and each instance is slightly different and has a different skill set.
- For example, Siri on Apple TV has a different set of skill domains than on the iPhone and the HomePod is incapable of doing some of the most basic things that are available on the iPhone.
- Furthermore, the third-party skills are all implemented on the device meaning that if the user enables Siri on the iPhone to call an Uber, all of his other devices will remain unable to do it.
- This is because SiriKit is a client-side SDK meaning that all the extensions run only on the client resulting in the limitation that I have outlined above.
- By contrast Google, Amazon and Baidu have implemented the intelligence and the third-party extensions in the cloud meaning that there is only one version sitting in the cloud that can be accessed from any enabled device.
- When a skill is enabled on one device, it is enabled on all of them giving complete uniformity of user experience.
- Google Assistant is now so good at this that if two devices are in range of the voice it will answer the request on the device that is most suited to delivering the answer or service requested.
- As voice becomes more important in controlling electronic devices, Siri’s fragmentation is going to lead to frustration and disappointment among users.
- We are already starting to see this in the HomePod where the consensus review opinion is: fantastic speaker, poor digital experience.
- Hence, this is a device that competes almost entirely on hardware which will work for a while and then cause real problems as competition catches up.
- Harmon Kardon, JBL, Sonos etc are the real competitors for this product and their flexibility around digital assistants and third-party services may well give them an edge.
- Some time ago, Apple put together its disparate AI research efforts but it seems to me that it has not gone far enough.
- It would appear that there is a central repository where the master code line of Siri is developed which is then checked out and customised by product teams to make it more applicable for the device they are creating.
- This results in fragmentation and inconsistency that wastes R&D resources and hurts the user experience.
- This is exactly the problem that contributed to Samsung’s and Motorola’s inability to mount a serious challenge to Nokia in feature phones 15 years ago.
- In Apple’s case I think that this still a reasonably straight forward fix by moving Siri and SiriKit completely to the cloud which would be a great feature to announce at WWDC this year.
- Failure to make this change will simply hand the advantage even more to Google, Baidu and Amazon who are far and away the leaders in this space.
- The question is whether Apple has the depth of character to make this change as it now has a massive business that is at the peak of its power and the requirement not to risk put this at risk will make it hesitate.
- This is exactly the hesitation that made innovation at Nokia very difficult and sank its smartphone business.
- This is why I think Apple must make this change, because if a high quality digital assistant becomes a requirement in the user’s purchase decision, Siri in its current form will start to play against users when purchasing digital ecosystem devices.
Reply to this post
Waymo’s lead shrinks to 5 to 1.
- Analysis of autonomous testing data submitted to the California DMV for 2017, reveals that Waymo is still out front, but GM Cruise has improved massively with Nissan holding onto its No.3 position.
- The best measure of an autonomous driving solution remains how often the driver has to take over to correct shortcomings in the autonomous driving software.
- Regulations in California require those that test in the state to submit this data but typically, they all submit it in different ways (see here).
- There are also different types of disengagement such as when the car is going to hit something (critical) or when the safety driver feels uncomfortable (ordinary).
- Furthermore, companies test their cars in different conditions meaning that the data can really only be used as indication.
- However, the contrast between the have’s and have-nots is so stark that meaningful conclusions can still be drawn by parsing the data submitted.
- In order of performance the data for 2017 shows:
- No. 1 Waymo (Google) is now only 5x better than its nearest rival compared to 8x in 2016.
- However, Waymo’s performance has only improved marginally with 5,596 miles per disengagement in 2017 compared to 5,128 in 2016.
- Waymo has still driven more miles than anyone else but its lead in miles driven has dropped from 155x its next rival to 2.8x.
- Curiously, Waymo drove 45% fewer miles in 2017 (353K) compared to 2016 (636K).
- No. 2 GM Cruise comes from nowhere to comfortably take the no.2 position with 125K miles driven with 105 disengagements (1190 miles / disengagement).
- This is 5x more than Waymo, but GM Cruise drove all of its miles in downtown San Francisco which it argues is the most difficult environment within which to operate an autonomous vehicle.
- I would argue that it is certainly the most complex but also the slowest, giving the computer much more thinking time than it has on suburban street or a highway.
- Either way, this is an impressive performance and for the first time, there is a credible challenge to Waymo’s dominance.
- No. 3 Nissan which is the other dark horse in this race with 5007 miles driven and 24 disengagements (209 miles per disengagement).
- Last place Mercedes which saw a meaningful deterioration in its performance during 2017.
- In 2017 Mercedes had nearly 1 disengagement for every mile driven compared to 2 per mile in 2016.
- It is possible that Mercedes decided to test in more challenging conditions in 2017 which caused the deterioration but regardless, it is the laggard in the race for an autonomous car driving solution.
- Those that drove the most miles (Waymo and GM Cruise) still had the best performance, again underlining that the key to artificial intelligence (the heart of all autonomous driving systems) remains the amount data collected (see here).
- Uber and Tesla have no data for 2017 as they did not test in California but given the distractions that Uber has suffered in 2017 combined with the ongoing trade secrets trial against Waymo, is likely to have meant that it has not improved much on its dreadful performance in 2016 (see here).
- I am certain that Waymo is the best because it began working on autonomous driving in 2009 (far earlier than anyone else) and in the last 2 years has also driven substantially more miles than anyone else.
- I continue to believe that there is not much point in rushing to get an autonomous driving solution to market as I still do not expect the market to be ready for autonomous vehicles much before 2028 (see here)
- Consequently, those rushing to market may find that they have a solution but no deployments.
- This could easily result in a number of viable solutions being available once the market is ready, making sitting on the sidelines for now a wise choice.
- I suspect that this is why Google and Waymo are currently pushing to do a number of large long-term deals for technology as their lead over the rest of the market has already passed its peak.
Reply to this post
HomePod delays have cost it dearly.
- The HomePod has only one advantage in a market that is rapidly becoming all about the algorithms with hardware being left to those that really know audio.
- The much awaited HomePod is now available for pre-order and will start shipping on February 9th
- At $399, it is going up against some of the very best that the audio industry has to offer.
- In 2017 the best digital assistants were only available in their own in-house hardware but this is changing rapidly as there is a plethora of third party products coming to market.
- Furthermore, Amazon Alexa, Baidu DuerOS and Google Assistant all have substantial advantages over Siri, which in my opinion, make them much better smart assistants to have in one’s house.
- Hence, I see the HomePod competing on two fronts and on the most important front, I see it being hopelessly outclassed.
- These are:
- Audio: There is no doubt that Apple can make excellent quality audio products.
- However, in this category, it is not alone.
- While I think it can comfortably hold its own in the mid to high end of the speaker market, I have doubts whether it can do so at 40% gross margins.
- This is because competing solely on audio quality, there will be just as good audio products available at lower prices.
- Digital Assistant. It is here that I think that Siri is hopelessly outclassed.
- My tests have consistently shown that both Google Assistant and Duer OS are much better products in their relevant markets. (Baidu DuerOS does not yet exist in English).
- Furthermore, when it comes to the smart home, Apple is hopelessly outclassed when it is compared to both Amazon Alexa and now, Google Home.
- The star of CES was Google Assistant and not just because Google bought almost every piece of advertising space that there was available.
- This was followed through on the show floor with almost every smart device manufacturer either already supporting Google Assistant or having it on the immediate roadmap.
- Just like 2017, Apple HomeKit was a no show and I saw just one product (a smart ceiling fan) that had support for Apple’s smart home offering.
- Consequently, I think that Siri is way behind in the smart home leaving Apple competing pretty much on audio quality alone.
- The real competitors for HomePod are the likes of Sonos, Sony, Harman Kardon and so on, all of whom are likely to make their product available with either (or both) Google Assistant or Amazon Alexa.
- This is why I see competition in this space moving away from audio quality and rapidly becoming all about the algorithms.
- This combined with Google closing the gap in the smart home, is what has lead me to reverse my position and to expect that it will be Google that ends up triumphant in this space (see here).
- Hence, from Apple perspective it appears that there will be equivalent sounding products with much better brains on the market at lower prices.
- I think only a small sliver the hardcore Apple fanbase is going to buy this product which is not enough to make it a real success in my opinion.
Reply to this post
Samsung should capitalise on Google’s failures.
- Samsung is making a lot of noise around its reimagining of the camera and AI on its new flagship Galaxy s9, but if it really wants differentiation in 2018, it would be wise to capitalise on Google’s failure to successfully launch its own products.
- The Google Pixel 2 and 2 XL have been a disaster not because they were bad products but because of a series of schoolboy errors (see here) that completely masked some superb smartphone innovation and kept volumes insignificant.
- Two examples of this innovation are:
- First, portrait mode: This feature is now common on dual camera devices as the difference between the foreground and the background can easily be calculated using two cameras.
- The background can then be blurred to produce the pleasing portrait effect.
- Google has managed to do this to a competitive grade using just one camera giving a substantial benefit (lower cost and more space) to any hardware maker that can deploy it.
- Second, machine vision: Google launched Google Lens on Pixel devices and while it still needs some work, it is clearly the best image recognition that is available on a smartphone today.
- For me, these were two of the most significant differentiators for Pixel devices but because Google bungled many other aspects of the hardware, they are effectively absent in the market today.
- Consequently, I see an opportunity for Samsung to maintain differentiation on new Google features by doing what Google seems incapable and of doing so at massive volume.
- If the Galaxy s9 launches with Google Assistant front and centre instead of the dreadful Bixby, Google Lens and with portrait mode on its camera rather than a variable aperture (which no one is likely to care about), then I think Samsung will have achieved differentiation for 2018.
- This will help it hold share and margins for this year as its competitors will unlikely to have these features before 2019.
- Granted, next year would see all of its competitors doing the same but this is exactly what has happened with the edgeless screen differentiation that it created in 2017.
- Consequently, I think that a Galaxy s9 that launches with these features front and centre rather than a variable aperture camera and awful AI that no one is going to use, will result in a better selling, higher priced product.
- At the end of the day, hardware is ancillary to Google whose objective in life is to generate as much traffic as possible within its services, providing it with greater opportunities for monetisation.
- In that context, hardware is merely the conduit for its services which is why I have never really understood its expensive obsession with making hardware as it has no hope of becoming like Apple.
- Hence, it would make a lot of sense for Google to allow Samsung to deeply embed the services and AI that it has previously made exclusive for Pixel devices.
- Making hardware is what Samsung really knows how to do and I think that the Samsung Galaxy flagships would be much better products as a result.
- Samsung would also have the option to stop making investments (like its AI and Bixby) that actually have a negative return and generate higher margins as a result. (I would argue that Bixby has done more harm than good to the s8.)
- The other manufacturers would of course be put out if they thought that they were being unfairly treated by Google, but I think that they nowhere else to go rendering their displeasure irrelevant.
- Sadly, I suspect that Samsung is going to continue to think that it can make a difference in AI meaning lower profits for shareholders.
- Samsung has run far farther than I thought it would and is now an investment decision taken on semiconductors and panels rather than phones.
- Hence, I continue to prefer Microsoft and Tencent and specifically Baidu for AI.
Reply to this post
Google shows the most progress at CES.
- Apart from automobiles, the prevailing theme at CES is one of boredom.
- Those looking for new trends have instead found the same stories as last year, albeit a bit more advanced than the last time that they were told.
- This steady progress is the real theme of CES as hype has been wound up to fever pitch levels over the last few years and now the time has come to begin delivering.
- The result is likely to be a steady year where autonomous vehicles inch closer to reality, where AR penetrates deeper into the enterprise but not where new trends rock the industry to its core.
- AI is going to continue to be a theme, but I think that 2018 will be another year of very slow development.
- Contrary to popular belief, RFM research has concluded that there has been very little progress in breaking down the really big problems of AI and without a new approach not much is going to change.
- The current favoured technique, backpropagation, was discovered in 1986 and took 26 years to start producing decent results.
- There are plenty of new techniques being worked on, but none of them have produced any results.
- Consequently, I am pretty certain that 2018 will be another year of small increments dressed up as a big advance.
- The one area where I have seen movement is in the smart home which I have discussed below.
- Developments within the smart home have been steady but the options for developers and functionality has improved markedly.
- This is because in addition to splashing Google Assistant over every available surface in Las Vegas, Google has been putting a lot of effort into pushing the assistant to developers.
- Over the last 3 days I have spoken with or passed by the stalls of over 100 companies making a smart gadget of some description.
- When I carried out this exercise last year everyone was supporting Amazon Alexa and almost no-one was supporting Google Assistant.
- This year, everyone is still supporting Alexa but they have also either already included Google Assistant or have put it on the immediate roadmap.
- This is a big change from 2017 and I think it substantially reduces the appeal of Amazon Alexa as Google Assistant remains a far better service.
- The one exception is shopping and in this function Google is hopelessly outclassed.
- However, Amazon is giving away the Dash Wand and I can see a scenario where users keep that stuck to the fridge for groceries and use Google Assistant for everything else.
- Google’s position in the smart home has not improved quite as much as I was expecting, but its improvement combined with the way that users are interacting with digital assistants, leads me to change my position.
- I think that Google now has the edge over Amazon is it has by far the better product and its native presence on smartphones means that it is dealing with far more inquiries than Amazon.
- Hence, I think that Google should be able to continue to improve the assistant relative to Amazon thereby steadily increasing its relative appeal over time.
- Furthermore, with a multitude of 3rd party products coming to market, this will no longer be a battle over hardware and sound quality but will become one fought in the ecosystem.
- Here, I continue to think that Google has Amazon soundly beaten, the results of which I expect to see over the coming 24 months.
Reply to this post
No Planes, no trains, only automobiles.
- By far the strongest theme at CES 2018 is the digitisation of the automobile with drones, VR and so on fading somewhat into the background.
- Byton is an electric car start-up run by Europeans (ex-BMW) but financed and headquartered in China.
- From the very first moment Byton had a tough hill climb as it was not so long ago that LeEco muddied sentiment towards Chinese EV start-ups with a glitzy event that was swiftly followed by a precipitous crash.
- Byton’s launch looked very similar (including the gaffs) but it was at the badly attended Q&A event that my scepticism was partially dispelled.
- Byton is not really about making an electric car as building one of these is going to be considerably easier than the internal combustion variety.
- Instead, Byton has focused its efforts on solving the fundamental problems that arise when one attempts to digitise the vehicle.
- These include addressing the issues that arise with traditional product cycles where the infotainment unit is 4 years out of date the minute it hits the market.
- RFM research (see here) has uncovered a series of fundamental problems with the way OEMs design and build cars and to its credit, Byton has come up with a credible solution to all of these issues.
- However, what it does not have is scale as its current $800m or so in funding (including the ongoing $350m B round) will take it to just 100,000 vehicles manufactured each year.
- This is tiny volumes meaning that the bigger OEMs will be able to offer a better specification of EV than Byton at a lower price.
- This means that Byton must differentiate itself on the digital experience that it offers which will be more difficult than it sounds.
- The true Digital Life of the automobile requires the combination of both the Digital Life Pie on the smartphone or tablet and the Digital Sensor Pie (see here).
- As an OEM Byton, will have full access to its Digital Sensor Pie but it will need to take that in context with the data generated by the Digital Life services of its users on their other devices.
- In China, this means dealing with the BATmen and in the West, this means Apple and Google.
- Byton is likely to go with an SDL like approach that uses an API to project smartphone apps and data onto the dashboard which requires the makers of those apps to support Byton.
- This will be a tough sell as the volumes are so low that convincing developers that Byton is worth the effort will not be easy.
- That being said, its innovative approach to the dashboard seems to be driving a lot of interest which might help it to drive support for its proposition.
- Byton has clearly put a lot of thought into its vision, but now it must execute and this is by far the hardest part.
- At Baidu World, it was clear what most of the audience was interested in as about one third of them left as soon as the presentation moved away from autonomous driving.
- I find this to be somewhat short sighted as when one is considering the importance of AI in Baidu’s future, I think its Duer OS platform is far more important.
- This is because Apollo is just about cars while Duer OS is about everything else and also includes cars.
- Baidu’s AI is centred around Baidu Brain (which does what it says on the tin) and Baidu Cloud which is where Baidu Brain lives.
- There are two main platforms based on Baidu Brain which are Apollo, its autonomous driving effort and Duer OS which is Baidu’s equivalent of Amazon Alexa, Cortana, Google Assistant and so on.
- Baidu has upgraded Apollo to version 2.0 (which I think is still pretty basic) as it can now drive on quiet urban roads bringing into line with what most of its Western competitors have been testing during 2017.
- It also announced real vehicles with Chinese partners Cherry and King Long (busses) and Los Angeles based mobility provider Access.
- Baidu also gave impressive growth figures for the traction of Duer OS and backed that up with a series of partner announcements and new devices.
- The combination of Baidu Chinese search graph and its leadership in Chinese AI makes it by far the leading digital assistant contender in China which was clear for all to see when looking at the demos.
- Baidu has also internalised the importance for cross device capability and its presence on phones, cars, TVs, speakers, lights, projectors, a mirror, a home robot and headphones is pretty comprehensive.
- This gives the user the ability to interact with his devices from any of the others which is something that the Western leaders have yet to properly demonstrate.
- However, Duer OSs’ international ambitions remain quite limited as it has launched a product only for Japan with Japanese language support but of English, Spanish and so on, there is no sign.
- Baidu has extended its leadership in AI in China but remains a fairly distant second behind Google which I still see as the global leader.
- That being said, Baidu’s recent travails have made it arguably the cheapest way to invest in AI in the global technology sector.
Reply to this post
RFM’s top 5 issues likely to prevail at CES and 2018.
- I expect Artificial Intelligence to get top billing again this year as the both the hype and the flow of capital show no sign of abating.
- Consequently, 2018 is likely to be a year of more feverish investment and hype making it more important than ever to separate real AI from those that are simply using statistics.
- The three big AI problems remain mainly unsolved (see here) and RFM has concluded that progress in 2017 was very slow despite plenty of noise being made to the contrary.
- I have no doubt that AI will become crucial for ecosystems trying to differentiate their Digital Life services from one another and the gap between the haves and the have nots is widening.
- Google has distanced itself further from its competitors and in my opinion remains by far the leader in this field.
Google vs. Amazon
- The battle of the digital assistants is likely to heat up this year and Google is clearly determined not to repeat the own goals of 2017 that allowed Amazon to dominate the market with an inferior product.
- Signs of this are everywhere at CES with the Las Vegas Conference Center and the casino monorail fully decked out with entreaties to use Google Assistant.
- Last year Google was nowhere to be seen at CES but this year I am hoping to see the results of its H2 2017 efforts through the inclusion of Google Assistant support by smaller developers in their smart home products and services.
- Although, Amazon dominates the market for devices it is capturing only a tiny fraction of the voice requests as 91% of users that interact via voice use a smartphone compared to 17% that use smart speakers (see here).
- Data is the life blood of AI and the data strongly suggests that Google is collecting far more than Amazon thereby ensuring that Google Assistant will continue to distance itself from Amazon Alexa in terms of ability.
- If Google manages to close the gap in smart home this year, I think that this will put Amazon on the back foot and on a trajectory towards losing the smart home to Google.
Smartphones – Bezels, folds and the race to the bottom.
- Bezel-less screens have become table stakes at the high end of the smartphone market meaning that 2018 will see this feature increasingly moving into the mid-range.
- Samsung created the bezel-less market just like it did for large screens and now it must now look for something else.
- The issue is that the Android user experience suffers from serious shortcomings compared to iOS meaning that it must offer othe features to compete at the iPhone price point.
- Samsung has had foldable screens for some considerable time but poor yield and a lack of interest has meant that they have never been launched.
- I have long seen the potential for foldable screens as a tablet form factor that can be folded away and slid into a pocket has the capacity to fundamentally alter both the tablet and laptop markets.
- 2018 may be the year that Samsung feels ready to finally launch this as its options in terms of maintaining differentiation in an increasingly crowded bezel-less market are looking thin.
- The theme of digitisation in the automobile is in full swing but 2018 is likely to be another year where hopes and dreams substantially outstrip reality.
- RFM’s analysis has shown (see here) that OEMs and tier 1s have not really digested the degree of change that is required for them to remain major players in their own industry.
- For example, by locking the development cycle of the infotainment unit to the rest of the vehicle, the industry has ensured that units for which users pay thousands of dollars for, are four to five years out of date and hopelessly outclassed by $150 smartphones.
- This combined with the almost universally awful user experience offered by automotive infotainment units puts the OEMs at risk of becoming also-rans in their own industry.
- It also leaves the door wide open for ambitious new-comers like Byton which has launched an EV and Digital Life experience which shows some signs of having been given a lot of thought to the experience issues plaguing the vehicle.
- With the exception of AR, very little is likely to change for both virtual reality and wearables in 2018 as the issues that hold them back remain unresolved.
- Wearables are still a solution looking for a problem while the health use case continues to be limited by the quality and reliability of the sensors that they use.
- Hence wearables will still be a recreational health and fitness market where users soon tire of their devices and consign them to cluttered junk drawers.
- I would still be placing all of my attention on the companies that are working on making medical grade sensors that are both cheap and reliable to wake this segment from its slumbers.
- I still see no real use case for VR beyond high-end gaming and events as the technical issues of cables, nausea and so on are still being worked on.
- This leaves AR which I think is going to have a good year in the enterprise.
- In the enterprise, the user experience matters less and the productivity use cases for AR in particular functions are numerous and demonstrable.
- This is why many AR companies have pivoted towards the enterprise leaving Magic Leap as one of the few that is left struggling along in consumer AR.
Reply to this post
January 5th 2018: Radio Free Mobile updates its research on Artificial Intelligence with the publication of Reality Bytes Issue No.2 – Practitioners and Pretenders.
RFM research subscribers will receive their copy directly by email.
Noise and speculation remain as high as ever but under the hood real progress remains painfully slow. The three goals of AI and Moravec’s paradox remain far from solved and may even cause progress to grind to a complete halt. Google DeepMind has produced what RFM considers to be real progress as its algorithms can thrash the best of the rest using far less than 1% of the resources. This has real implications for AI running in battery powered devices. Google Assistant has closed a lot of the gap to Amazon Alexa while Facebook will suffer financial pain from its weakness in AI.
- Three goals & Moravec’s paradox. Progress has been slow with the big issues still unsolved. These are: 1) the ability to train AIs using much less data than today, 2) the creation of an AI that can take what it has learned from one task and apply it to another and 3) the creation of AI that can build its own models rather than relying on humans to do it. Moravec’s paradox refers to the fact that tasks that are hard for humans are easy for machines and visa versa. Progress remains very slow even by the best in the field.
- Practitioners vs. pretenders. RFM continues to separate the practitioners from the pretenders by assessing understanding of and focus upon the three goals of AI as well as Moravec’s paradox. The practitioners are the ones that are likely to create the algorithms that result in long-term differentiation and profitability in digital ecosystems.
- Dead end? Almost all AI is based on a technique called backpropagation which has been around since 1986. It requires vast amounts of data and is poorly understood. This raises the possibility that progress grinds to a halt until a new technique is developed. This would lead to commoditisation as the laggards would have time to catch up with the leaders.
- Google & Deep Mind have made more noise than progress in 2017, but remain the real leaders in this field. RFM thinks DeepMind’s biggest achievement is the creation of an algorithm that can thrash anything and anyone at Go, Chess and Shogi but uses less than 1% of the resources. This has substantial implications for running AIs in smartphones.
- Digital Assistants. Google Assistant has made up a lot of ground on Alexa, leaving Amazon’s only edge as shopping via Alexa. This will be very difficult for Google to compete against but given its superiority everywhere else, RFM sees a possibility that Amazon gets squeezed out.
- Facebook’s weakness in AI will hurt financial performance as it has to recruit humans to do work that is much better done by machines. The result is likely to be only a minor improvement and a meaningful decline in profitability. AI remains the key risk to Facebook’s ambitions.