I got a Mac Studio and here are 5 reasons why you (probably) do not need it

Apple announced the Mac Studio during their March 2022 “Peek Performance” event alongside the M1 Ultra SoC, designed for creative professionals who need a lot of I/O and processing power.

I ordered one for myself on 17 April 2022 and it arrived on 7 July 2022. The specifications are as follows:

  • M1 Ultra (20-core CPU, 48-core GPU, 32-core Neural Engine)
  • 64 GB RAM
  • 2TB SSD

Originally I had no intention of getting the Mac Studio given my use case. Even though I’m mostly desk-bound, I did like the idea of having a versatile machine that I can also bring with me. Therefore, my plan was to get the M1 Max MacBook Pro.

However, there had been several issues with using a MacBook as a desktop that got me to re-think.

I have been dealing with bloated batteries on my constantly plugged-in MacBooks for long enough that I decided not to go through that again.

Secondly, laptops do not necessarily have very good thermal dissipation and that killed the components faster; it is especially so in Singapore where the ambient temperature are averaging 30-32 degrees Celsius. Constant exposure to such heat also contributes to battery swelling.

Now that I have the Mac Studio and have been using it for a few weeks now, I have a better perspective of who the machine is for.

Here are 5 reasons why you probably don’t need it.

The high price tag

In the eyes of most people, Apple products have never been cheap. There are cheap Windows PC and laptops that could perform the same if not better.

Its base M1 Max model (10-core CPU, 24-core GPU, 16-core Neural Engine, 32GB RAM, 512GB SSD) cost a whopping SG$2899. For similar price, you could get a rather powerful desktop PC (e.g. this or this ) from Aftershock that allows you to do the standard administrative and/or office task, browse the web, watch videos and video games.

If you are getting the base M1 Ultra model (20-core CPU, 48-core GPU, 32-core Neural Engine, 64GB RAM, 1TB SSD), it will set you back SG$5899. For that kind of money, you can get a very powerful gaming PC with Core i9, 32GB RAM and a RTX 3080 that allows you to do everything the Mac can do and provide you with sufficient computing power for the next 3 to 4 years.

Untapped processing power

Based on the performance benchmark and review done on Apple M1 series of SoC, the results shows that it matches, if not beat, vast majority of system running Intel or AMD CPU while consuming much lesser power.

With that in mind, the Mac Studio is designed for professionals who need that kind of computing power such as those involved in video production, music production or crunching through large datasets.

For most of us using our computers for general tasks such as working with documents, browsing the web, online shopping and/or watching videos, the cost/benefit ratio of getting a Mac Studio is not justifiable.

For all the tasks mentioned, the M1 chip found the MacBook Air is more than enough for everything mentioned. And the cheapest M1 MacBoon Air from Apple is SG$1500. Buying from elsewhere could give you an even lower price.

Even software engineers don’t need all that power provided by Mac Studio. And if they need servers to run their applications, they could easily spin one up on Amazon, Google or Microsoft’s Cloud Platform. Even if they need to run docker containers on their development machine, the 14inch MacBook Pro with M1 Pro (10-core CPU, 16-core GPU) and 32GB ram would suffice.

You are always on the go

The Mac Studio is a desktop computer. If your work requires you to be on the move often, the Mac Studio does not make any sense. A MacBook Air/Pro will make a better option.

Yes, you can bring the Mac Studio with you as it is compact enough but you will also need to bring the display with you. And you will also need to find a place with a power outlet.

No display or peripherals

When you buy the Mac Studio, all you get is the computer. You have to get keyboard, mouse and monitor separately.

If you prefer to have something that comes together in a single package like a laptop, then this machine is not for you. If you have severe space constraint that prevent you from playing host to a monitor, keyboard and mouse, this machine is also not for you.


Even with the issues stemming from using a MacBook Pro as a desktop replacement, they still don’t make a compelling case for me to get such an expensive machine. In fact, I will not recommend it for vast majority of people.

So, what was my reason?

There are two reasons why I got the Mac Studio:

  • I can afford it
  • I want the greatest and biggest that is within my budget

What is Wi-Fi 6 and why it is a game changer?

Wi-Fi 6 is the consumer friendly name for the IEEE 802.11ax, which is the current generation Wi-Fi specification standard and the successor to Wi-Fi 5, also known as IEEE 802.11ac.

This new standard comes with various improvements to efficiency and throughput. It is also backwards compatible with your older devices since it supports both 2.4 Ghz and 5 Ghz bands.

But, before we can dive deep into explaining what is Wi-Fi 6 and advantages it bring to the table, we need to be on the same page regarding certain terminologies. We shall also setup some analogies that we will use to explain technical details in simpler terms.

To connect to the internet, we will need to install either a modem or gateway provided to us by internet service providers. If it is the former, a router typically is connected to the modem to provide additional connection points such as ethernet ports and Wi-Fi for our devices to connect to in order to access the internet. If a gateway is used instead, the gateway itself is also a router and could also provide the same functionality as a standalone router.

With that in mind, let us imagine the router to be a large warehouse in a middle of a small city. It has a number of gates or main doors that represent antennas. In addition, there are buildings (or clients, if you will), which represents devices such as smartphones and laptops. Then, there are also couriers who represent radio waves. These couriers are the middlemen between the warehouse and the various buildings, and are responsible for transporting packages to the mall and back to their respective clients. These packages contain things, which are pieces of data, that the building need.

It is fast

Wi-Fi 6 has a theoretical maximum speed of 9.6 Gbps (gigabits per second) or 1.2 gigabyte per seconds. That is almost 2.6 times more than Wi-Fi 5. To put things in perspective, it is possible to transfer a 100GB 4K blu-ray video in just under two minutes with such speed.

There are three key technological updates that contribute to Wi-Fi 6’s improved performance over the previous generation: MU-MIMO, OFDMA and 1024-QAM


MU-MIMO stands for “multi-user, multiple input, multiple output”. It is a technology that help increase the number of antennas a router has and could be found implemented in most Wave 2 Wi-Fi 5 (IEEE 802.11ac) routers and devices.

Wave 1 Wi-Fi 5 and earlier routers could only communicate with one device one at a time and the others have to wait until it is their turn.

With MU-MIMO release for Wave 2 routers, they could communicate with up to four devices simultaneously.

The caveat here is that, the location of each client device matters. If two or marke devices are in the same general location, they will still need to wait in line to communicate with the router.

And you might wonder how does this technology help to improve performance?

Let us assume that a router will be placed in the center of the house or a room. Then, if there are multiple devices around that require Wi-Fi access, they could access the router via the nearest antenna pointed in their respective general direction. If two or more devices are in the same general location, then these devices will share the same antenna and need to wait for their turn.

For those who have a hard time understanding that, let us go back to the analogy we created earlier.

When the warehouse is first built, which represent the earlier version of Wi-Fi, there is only one gate. In order to enter or exit, couriers need to queue up by the gate and wait for their turn before they can proceed to either deliver or collect their respective packages.

Then, with MU-MIMO technology for Wave 2 Wi-Fi 5, the warehouse undergoes a renovation to have four gates installed according to the points on a compass instead of just the one. Now with four gates, more couriers can enter and exit albeit from four different directions at any given point in time. With this, the buildings are now able to get their requests serviced faster.

But if a particular gate has a long queue of couriers, they will still need to wait for their turn to enter or exit unless they choose to use other gates.

And for Wi-Fi 6, MU-MIMO is upgraded to support up to eight devices at the same time.

Going back to the warehouse example, renovation is done to add four more gates, for a total of eight. More couriers now can enter and exit the warehouse from eight directions at any given point in time, therefore servicing even more buildings.


Orthogonal frequency division multiple access (OFDMA)is an extension of Orthogonal frequency division multiplexing (OFDM) technology.

But, before we can dive deeper into OFDMA, we must first understand what is OFDM.

OFDM is a technology that takes a radio channel such as the 20 Mhz channel, which is often used for Wi-Fi, and sets a number of sub-carriers instead of having just one carrier. For Wi-Fi 5, 52 sub-carriers can be created from a single 20 Mhz channel using this technology while Wi-Fi 6 take that to the next level and can create 234 sub-carriers.

To the uninitiated, a carrier is a modulated wave that convey information.

Each sub-carrier is then modulated independently and simultaneously to form symbols, which are waveforms that represent information or data. And these symbols are separated in time by guard intervals to prevent interference cause by neighbouring sub-carriers. Finally, a single transmission to the receiving device will consist of a number of these simultaneous symbols spanning the sub-carriers.

The receiving device is able to track all these sub-carriers simultaneously and extract data from each sub-carrier independently. This contributes to the increase in Wi-Fi performance since multiple pieces of data are transmitted at the same time.

However, the problem with OFDM is that any device is free to transmit a signal whenever they are ready, creating a first-come-first serve situation. This may work in a home-setting where there are not a lot of devices but does not work in a high-density area such as a stadium and shopping mall. Too many devices will be fighting for a chance to send and receive data from the router, resulting in an inefficient use of the router.

For those who have trouble following the above explanation, let us go back to the warehouse example we used earlier.

OFDM can be thought of as multiple couriers who are dispatched to a specific building but at slightly different timings. They travelled on the same road but on a different lane, and each of them carries a different part of the final data. Once they arrived at their destination, the building’s manager will collect the different parts of the data and began the reconstruction process. Even if some of the couriers got lost, the manager is still able to reconstruct the data because each courier carries a nice little clipboard containing a detailed description of the content and its relation to the other couriers.

However, there could be couriers serving other buildings traveling on the same road. Some of these couriers could end up on a lane where other couriers are on. This can lead to a scenario where the couriers would fight amongst themselves in order gain access the warehouse first. During the fight, packages will be lost and when that happens, the buildings need to dispatch couriers with the same packages again in an attempt to gain access to the warehouse.

This is where OFDMA comes in.

OFDMA technology solves the network contention issue by grouping the sub-carriers into Resource Units (RU) to service one or more clients depending on their needs.

Therefore, if there is a client that need higher bandwidth because of the data it is downloading, then all the sub-carriers can be grouped as one Resource Unit to give the client the full bandwidth of the channel. Similarly, if multiple clients in the same area need a fraction of the bandwidth because of their small data requirement, then multiple Resource Units can be created to serve all of the clients.

And the change in the RU configuration is also done in real time, therefore enabling a consistent efficient use of the available network bandwidth.

To the layman, it is the equivalent of upgrading the warehouse to dispatch one truck per lane to serve a group of buildings that are close together. The purpose of the truck would be to carry as many couriers as it can hold and transport them to the group of buildings where they could then drop or pick up the packages.

When does the warehouse dispatches the trucks and decides whether the couriers on board all serve the same or different building are dependent on the requirements such as the service type, package size and total number of packages.

With that, the odds of couriers losing their packages and having to restart the transport process again, which is an overhead, is reduced. Furthermore, the different buildings (clients) in the same area get an equal amount of attention from the warehouse.


Wi-Fi 6 improves on the amount of data transmitted per signal, allowing improvement in speed by up to 30%. This means that you can stream bandwidth hungry content such as 4K video with further reduction in loading times and have a smooth viewing experience.

But before we could dive deeper into how Wi-Fi 6 achieve this, we need to understand how Wi-Fi work in general.

Wi-Fi works by using radio waves. To transmit data so that the receiving device understand it, the sender need to modulate the signal to represent bits of binary code. This type of modulation is known as “Quadrature amplitude modulation” or QAM for short.

The better a device is at modulation, the more information it can transmit each time.

For example, a 2-QAM device means it capable of transmitting one bit (1 or 0) of information each time because it can modulate the signal in one of the two ways. A 4-QAM device can transmit 2 bits (00, 01, 10, 11) of information each time because it can modulate a signal four different ways.

With that in mind, current generation of Wi-Fi 5 devices are 256-QAM, which means eight bits of information can be transmitted each time. This is why most of us today do not spend a lot of time waiting for video to load and buffer. With Wi-Fi 6, devices are able to do 1024-QAM, which means 10 bits of information can now be transmitted each time.

To explain QAM much simply, let us go back our warehouse example.

2-QAM is the equivalent of the courier only having one hand. They can either carry one bag or nothing at all. 4-QAM give them another hand, so now they can carry up to two packages. 256-QAM for Wi-Fi 5 is the equivalent of giving a courier four pairs of hands, thereby enabling them to carry up to eight packages. With the upgrade to 1024-QAM for Wi-Fi 6, each courier now has five pairs of hands to carry up to ten packages.

On the surface it may not look like much. However, if a request is for a large amount of data such as those typically found during 4K movie streaming, having the ability to transfer more data per trip will mean less trips needed to download the full content. After all, less trips equals more time saving.

It could improve battery life

Other than being faster, Wi-Fi 6 also comes with a new feature call Target Wake Time. This allows certified Wi-Fi 6 routers to schedule check-in times with connected devices.

With scheduling, devices only activate their antennas at the right time instead of having to keep their antennas powered on to transmit or search for signals for an extended period of time, which can consume quite a fair amount of power.

For devices such as laptops or desktops which are connected to a power source and do need persistent internet connection, this feature may not be useful. But for IoT devices it could be a world of difference since they may not have access to consistent power and probably run on batteries.

It has better security

Since 2004, the Wi-Fi security revolves around WPA2. It is a protocol that encrypts the communication session between the router and the client device so that they could exchange information safely and privately.

WPA2 was considered to be very secure until 2017 when a weakness in the protocol was discovered that made it possible for attackers in range of the Wi-Fi router to steal sensitive information.

The Wi-Fi Alliance announced WPA3 in 2018 to be the replacement. WPA3 replaces the need for the 4-way handshake to authenticate a client in WPA2 with another method called Simultaneous Authentication of Equals (SAE).

SAE is a proven zero-knowledge method to establish a secret shared key that both the client and the router will use to generate the session key to encrypt and decrypt Wi-Fi transmissions. If another client wishes to connect to the network, the client will established its own secret shared key with the router.

The other important feature of WPA3 is Forward Secrecy, which is an indirect effect of implementing SAE. This ensure that even if an attacker managed to capture the encrypted Wi-Fi transmissions and then crack the session key, older data continue to remain unaccessible as the keys used to encrypt those data will be different.

WPA3 is optional for existing devices and many device manufacturers may choose not to patch these products via firmware update. But in order for these manufacturers to market their devices to be Wi-Fi 6 certified, the Wi-Fi Alliance mandated that WPA3 be implemented. Therefore, we can be sure that Wi-Fi 6 will be more secure.

Other than improvements made to the WPA protocol, the security and privacy of open Wi-Fi networks such as those we find in cafes, shopping malls and stadiums are also improved. Wi-Fi 6 will see the implementation of Opportunistic Wireless Encryption (OWE).

OWE is a security technique that is similar to SAE to encrypt the transmission channel between the device and the router but without the need for authentication. The established shared key is only known to the client device and the router.

Although it is not as secure since there is no way to tell who is connected to what, it is more secure than connecting to a public Wi-Fi secured by WPA2 and using a the pre-shared password, or connecting to a completely open Wi-Fi network.

More than one million WordPress sites attacked over the weekend of late May 2020

WordPress throughout its history has always found itself appearing in the news for its security vulnerabilities. The most recent vulnerability incident with WordPress is with a plugin call Page Builder by SiteOrigin.

Attackers mount a campaign over the weekend of 29 – 31 May against more than one million WordPress sites in an attempt to download wp-config.php, a file critical to all WordPress installations. This file contains sensitive information such as database credentials, connection information as well as unique authentication salt and keys. Therefore, anyone with access to the file could gain access to the database where the site content and users are stored.

To download that file, the attackers targeted cross-site scripting (XSS) vulnerabilities found in older plugins or themes that allow files to be downloaded or exported.

The attacks came from more than 20,000 IP addresses, which were also implicated in a previous attack that happened earlier in May 2020 used by the same threat actor.

The earlier attack targetted a different set of XSS vulnerabilities with the intention of having visitors redirected to malvertising sites. This set of vulnerabilities were found in plugins that have mostly been patched or plugins that have been removed from the WordPress plugin repository. Below is the list of plugins and their respective vulnerabilities that were popular with the attackers.

  • Easy2Map plugin — Removed from WordPress plugin repository due to XSS vulnerability
  • Blog Designer — XSS vulnerability that was patched in 2019
  • WP GDPR Compliance — Options update vulnerability that was patched in late 2018
  • Total Donations — Removed from Envato Marketplace permanently. It had a critical options update vulnerability.
  • Newspaper theme — XSS vulnerability that was patch in 2016.

The good news is that WordPress site owners who uses Wordfence are protected. According to Ram Gall at Wordfence, the Wordfence firewall blocked over 130 million attacks intended on harvesting database credentials.

How do you know if you were attacked?

The attack should be logged. You could look for any log entries that contain wp-config.php in the query string with the HTTP response code 200.

Below are the top 10 IP addresses used for this attack campaign.


What should you do next?

WordPress sites running Wordfence are protected from the attack. For the other users, you should change the database password and the unique authentication keys and salt immediately if you believe you are compromised.

The reason is simple.

WordPress servers that have been configured to allow remote database access could easily allow an attacker with the database credentials to add an administrative user, extract sensitive data or delete the site. Even if remote database access is not enabled, an attacker who knows the authentication keys and salts could bypass other security mechanisms that protect your site more easily.

And what if you are not comfortable making changes mentioned above?

Then you should contact your host or service provider since changing the database password without updating the wp-config.php file can render your site offline temporarily.

Last but not least, you should also update any plugins and themes. You may also want to consider changing the plugins or themes if these are no longer maintained by the original developers.

This article uses material from Wordfence.

A personal opinion on writing on a touchscreen device

Writing is still writing no matter the platform. It’s all about getting the words out, to give them a physical form be it on the screen or on paper. You can write on a piece of paper using a pen. You can write using your smartphone. You can write on your laptop or a desktop computer.

But what I have discovered is that writing on a touchscreen just feel weird and difficult. Some people no doubt won’t have any problems. It’s just not the thing for me.

I got the iPhone X. With its 5.8 inch, nearly edge-to-edge display, it’s way bigger than the iPhone 6s and 7 plus display I used in the past. That means with apps like iA Writer, I can see way more of the text with the keyboard below. The Super Retina HD display meant that text are sharp and clear. Writing on that device had been a joy.

Yet, whenever I tried to write long form, like a short story, my fingers do get really tired from attempting to hit the keys. My fingers are rather fat. Combine that with hyperhidrosis, it means either wrong keys are pressed and I need to hit delete or that the key presses aren’t registered like it should. It slows down my writing by a lot, which is irritating in a way if your thoughts is faster than the words appearing on the screen.

The other issue I have with typing on a touchscreen was the lack of tactile feedback. This is one of the reason why I prefer to write using a keyboard. The sound my finger hitting the keys and the clacky feel when you press the key just feels so good. I know you could enable haptic feedback on the phone such that every key pressed will give you a vibration. But that vibration is missing when you set the phone to silent mode via that switch. Not only that, vibration requires the motors in the phone to work hard and cause faster battery drainage. For the iPhone X, that vibration mode is no more and what you get is simulated keyboard clicks, something that you won’t hear if your phone is on permanent silent mode.

The third issue I have is having to deal with the weight of the device while typing. I know smartphones are small and consider rather light. After all you carry it in your pockets everyday. But it does become heavy when you are holding it in your hands for long period of time as you type. And that particular use case happens quite often if you are writing a long article, an essay or stories. Notes taking is fine actually because those are short burst action and probably won’t be doing it over 1 or 2 hours.

So those three reasons are why I will always prefer to write on a keyboard. And in order to do writings on the go, a portable typing machine is needed. Thus, I decided to reuse the 13inch MacBook Pro (2015) that was in storage. The 15inch MacBook Pro that I’m currently using is just a tad bigger and heavier than what I would like. You know what? Without the keyboard cover, typing on that classic chiclet keyboard is rather delightful. I could type equally fast on it.

And now I’m curious about what’s the primary device that you use to write everyday? And why.

Billions of Wi-Fi devices are vulnerable to eavesdropping due to

At the RSA security conference, security researchers announced that there is a Wi-Fi vulnerability that affects billions of devices. This vulnerability allows nearby attackers to decrypt sensitive data that are sent over the air.

Eset, the security company that discovered vulnerability, named it Kr00k and it is tracked as CVE-2019-15126. Kr00k affects the Wi-Fi chips made by Cypress Semiconductor and Broadcom. FullMAC WLAN chips from both companies are especially affected according to Eset. These chips are used in billions of devices and some of the devices include the following:

  • iPhones
  • iPad
  • Apple Macs
  • Amazon Echos
  • Amazone Kindles
  • Android devices
  • Raspberry Pi 3
  • Wi-Fi routers from Asus and Huawei

Most of the affected devices have patches made available by manufacturers but it is not clear how many of them installed the patches. Routers have the biggest concern because they often go unpatched indefinitely.

How does the vulnerability work?

When a wireless device disassociate from a wireless access point, unsent data frames will be placed in a transmit buffer and then sent over the air. Kr00k exploits this weakness. If either the device or the wireless access point has the flaw, these data frames will be encrypted with a key consisting of all zeroes instead of the session key negotiated earlier by the wireless device and the wireless access point. The use of a key consisting of all zeroes to encrypt data is equivalent to having no key.

The following diagram from ArsTechnica shows what would happen when a device disassociate from a wireless access point if either one is vulnerable.

A disassociation typically happens when a client device roams from one Wi-Fi access point to another, encounters signal interference, or has its Wi-Fi turned off. Hackers within range of a vulnerable device or access point can easily trigger this vulnerability by sending disassociation frames since they are not authenticated. From there, hackers could then capture and decrypt the transmitted data. They could trigger multiple disassociation to improve their chance of obtaining useful data.

The following diagram from ArsTechnica shows how the attack would happen.

What are the devices affected?

Eset researchers identified a variety of mobile devices that are vulnerable, including:

  • Amazon Echo 2nd gen
  • Amazon Kindle 8th gen
  • Apple iPad mini 2
  • Apple iPhone 6, 6S, 8, XR
  • Apple MacBook Air Retina 13-inch 2018
  • Google Nexus 5
  • Google Nexus 6
  • Google Nexus 6S
  • Raspberry Pi 3
  • Samsung Galaxy S4 GT-I9505
  • Samsung Galaxy S8
  • Xiaomi Redmi 3S

In addition, the following routers are also vulnerable:

  • Asus RT-N12
  • Huawei B612S-25d
  • Huawei EchoLife HG8245H
  • Huawei E5577Cs-321

The researchers also tested Wi-Fi chips from other manufacturers, including Qualcomm, Realtek, Ralink, and Mediatek and did not find any evident of them being vulnerable. However, since it was impossible to test all devices, it is possible that other devices using Cypress and Broadcom chips are affected.

For Apple, the vulnerabilities were patched in October 2019 as part of macOS Catalina 10.15.1, Security Update 2019-001 for macOS Mojave, Security Update 2019-006 for macOS High Sierra, iOS 13.2 and iPadOS 13.2 More information on the patches could be found here for macOS and here for iOS and iPadOS.

Amazon also state that Amazon Echo and Kindle devices listed in the security research have received automatic security update over the internet in a separate statement to ArsTechnica.

An opinion on improving voice user interface while ensuring privacy

Voice user interface is going to be one of the ways we interact with our devices as we go about our daily lives. It is just a very intuitive way for us because we communicate primarily via voice with text and images to complement.

But there still are various problems that need people to work on them to improve the overall experience. One of it is related to how the AI behind voice user interface can interact with us more naturally, like how we interact with fellow human beings.

This article written by Cheryl Platz got me thinking about that. It also covered a little on privacy and why it is a contributing factor that make it difficult for current generation of AIs to speak more naturally and understand the context when we speak. Unless, companies don’t give a shit about our privacy and start collecting even more data.

In this article, I am going to share what I think could help improve the AI and ensure user privacy.

Current Implementations and Limitations

What an AI needs to be better at understanding and responding in ways most useful to us are processing power, a good neural network that allows it to self-learn, and a database to store and retrieve whatever it has learnt.

The cloud is the best way for an AI to gain access to a huge amount of processing power and large enough database. Companies like Amazon and Microsoft offer cloud computing and storage services via their AWS and Azure platform respectively at very low cost. Even Google offers such services via their Compute Engine.

The problem with the cloud is reduced level of confidence when privacy is involved. Anything you store up there is vulnerable and available for wholesale retrieval through security flaws or misconfigurations. Companies could choose to encrypt those data via end-to-end encryption to help with protect user’s privacy but the problem is the master keys are owned by said companies. They could decrypt those data whenever they want.

Or you could do it like what Apple did with Siri, storing data locally, and use Differential Privacy to help ensure anonymity but it reduces the AI capabilities because it doesn’t have access to sufficient amount of personal data. Two, Siri runs on devices like Apple Watch, iPhones and iPads, which could be a problem when it comes to processing and compute capabilities, and having enough information to understand the user.

Although those devices have more processing power than room-sized mainframes from decades ago, it’s still not enough, energy-efficiency and capability wise, to handle highly complex neural networks for better experience with voice user interfaces.

Apple did try to change that with its A11 Bionic SoC that has a neural engine. Companies like Qualcomm, Imagination Technologies, and even NVIDIA are also contributing to increase local processing power with energy efficiency for AI through their respective CPU and GPU products.

Possible Solution

The work on the hardware by companies should continue so that there will be even more powerful and energy efficient processors for AI to use.

In addition to that, what we need is a standard, wireless-based protocol (maybe bluetooth) for the AI on our devices, irrespective of companies, to talk to each other when they are near to each other and in our home network. This way, the AI on each of those devices can share information and perform distributed computing, thereby improving its accuracy, overall understanding of the user, and respond accordingly.

A common software kernel is also necessary to provide different implementation of neural network a standardized way of doing distributed computing efficiently and effectively.

So now, imagine Siri talking to Alexa, Google Assistant or even Cortana via this protocol and vice versa.

Taking privacy into account, information exchanged via this protocol should be encrypted by default with keys owned only by the user. Any data created or stored should only reside on device also encrypted and nowhere else. Taking a page out of Apple’s playbook, the generated keys should come from some kind of hardware-based “Secure Enclave”.

To further improve the neural network, Differential Privacy should be applied on any query or information sent by the AI to the cloud for processing.


The above is really just my personal thought of how current the AIs powering voice user interfaces can be improved.

At the end, it’s really up to the companies to decide if they want to come together and improve all our lives taking into account our privacy and security.