Considerations for a More Ethical Approach to Data in AI: On Data Representation and Infrastructure

Data shapes the development of Artificial Intelligence (AI) as we currently know it, and for many years centralized networking infrastructures have dominated both the sourcing and subsequent use of such data. Research suggests that centralized approaches result in poor representation, and as AI is now integrated more in daily life, there is a need for efforts to improve on this. The AI research community has begun to explore managing data infrastructures more democratically, finding that decentralized networking allows for more transparency which can alleviate core ethical concerns, such as selection-bias. With this in mind, herein, we present a mini-survey framed around data representation and data infrastructures in AI. We outline four key considerations (auditing, benchmarking, confidence and trust, explainability and interpretability) as they pertain to data-driven AI, and propose that reflection of them, along with improved interdisciplinary discussion may aid the mitigation of data-based AI ethical concerns, and ultimately improve individual wellbeing when interacting with AI.


INTRODUCTION
Artificial intelligence (AI) in its current form relies heavily on large quantities of data (Yavuz, 2019), and data-driven Deep Neural Networks (DNNs) have prompted fast-paced development of AI (Greene, 2020). Currently, the research community is under great strain to keep up with the potential ethical concerns which arise as a result of this (Naughton, 2019). Within the AI community such ethical concerns can require quite some disentanglement (Allen et al., 2006), and it is not until recently that AI-based research groups have begun to provide public manifestos concerning the ethics of AI, e.g., Google's DeepMind, and the Partnership AI. 1 The Ethics of AI (Boddington, 2017) is now an essential topic for researchers, both internal and external, to core-machine learning and differs from Machine Ethics (Baum et al., 2018). The latter refers to giving conscious ethical based decision-making power to machines. The Ethics of AI, although somewhat informing Machine Ethics, refers more broadly to decisions made by researchers and covers issues of diversity and representation, e.g., to avoid discrimination (Zliobaite, 2015) or inherent latent biases (van Otterlo, 2018). Herein, our discussion focuses on topics relating to the Ethics of AI unless otherwise stated.
There has been recent research which shows promise for improved data learning from smaller quantities ("merely a few minutes") of data . However, machine learning algorithms developed for AI commonly require substantial quantities of data (Schneider, 2020). In this regard, Big Data ethics for AI algorithms are an expanding discussion point (Berendt et al., 2015;Mittelstadt and Floridi, 2016). Crowdsourcing (i.e., data gathered from large amounts of paid or unpaid individuals via the internet), is one approach to collect such quantities of data. However, ethical concerns including worker exploitation (Schlagwein et al., 2019), may have implications on the validity of the data. Additionally researchers utilize in-the-wild internet sources, e.g., YouTube (Abu-El-Haija et al., 2016) or Twitter (Beach, 2019), and apply unsupervised labeling methods (Jan, 2020). However, in Parikh et al. (2019), the authors describe how approaches for automated collection and labeling can result in the propagation of historical and social biases (Osoba and Welser IV, 2017). In the health domain, such bias could have serious consequences, leading to misdiagnosis or incorrect treatment plans (Mehrabi et al., 2019).
One method to avoid bias in AI is through the acquisition of diverse data sources (Demchenko et al., 2013). With Veracity (i.e., habitual truthfulness) being one of the 5 Vs (e.g., Velocity, Volume, Value, Variety and Veracity) for defining truly Big Data (Khan et al., 2019). However, big data is commonly, stored in centralized infrastructures which limit transparency, and democratic, decentralized (i.e., peer-to-peer blockchain-based) approaches are becoming prevalent (Luo et al., 2019).
Centralized data storage can be efficient and beneficial to the "central" body to which the infrastructure belongs. However, it is precisely this factor amongst others (i.e., proprietary modeling of underrepresented data) that are problematic (Ferrer et al., 2019).
Furthermore, centralized platforms limit the access and knowledge that data providers receive. The General Data Protection Regulation (GDPR) was established within the European Union (The-European-Commission, 2019) to partly tackle this. GDPR is a set of regulations of which the core goal is to protect the data of individuals that are utilized by third parties. In its current form, GDPR promotes a centralized approach, supporting what are known as commercial governance platforms. These platforms control restrictions to employees based on a data providers request but primarily function as a centralized repository. In essence, GDPR meant that companies needed to re-ask for data-consent more transparently. However, the "terms of agreement" certificate remains the basis, and 90 % of users are known to ignore its detail (Deloitte, 2016).
As a counter approach to the centralized storage of data, for some time researchers have proposed the need for a decentralized (cf. Figure 1) networking in which individual data is more easily protected (i.e., there is no "single point" of failure). In this infrastructure, individuals have more agency concerning the use of their data (Kahani and Beadle, 1997). Primarily, individuals choose to access parts of a network rather than its entirety. On a large scale, this paradigm would remove the known biases of centralized networks, as targeted collection, for example, would be less accessible by companies and sources of the data more complex to identify. In this way, various encryption algorithms, including homomorphic encryption (a method which allows for data processing while encrypted), or data masking, are being integrated within decentralized networks, allowing for identity preservation (Setia et al., 2019). Federated Learning (FL) (Hu et al., 2019), is one approach which can be applied to decentralized networks to improve privacy (Marnau, 2019). In FL, weights are passed from the host device and updated locally, instead of raw data leaving a device (Yang et al., 2019).
With these topics in mind, in this contribution, we aim to outline core ethical considerations, which relate to data and the ethics of AI. Our focus remains on the ethics of data representation and data infrastructure, particularly selection-bias and decentralization. We chose these topics due to their common pairing in the literature. A regular talkingpoint in machine learning is selection-bias and a networking infrastructure which may help to more transparently observe this is decentralization (Swan, 2015;Montes and Goertzel, 2019).
Our contribution is structured as follows; firstly we shortly define key terminology used throughout the manuscript in section 2, followed by a brief background and overview of the core themes as they pertain to AI in section 3. We then introduce our ethical data considerations in section 4 providing specific definitions and general ethical concerns. Following this in section 5, we connect these ethical considerations more closely with data representation and infrastructure, and in turn, outline technical approaches which help reduce the aforementioned ethical concerns. Finally, we offer concluding remarks in section 7.

TERMINOLOGY
There are a variety of core terms which are used throughout this manuscript which may have a dual meaning in the machine learning community. For this reason, we first define here three core terms, ethics, bias, and decentralization used within our discussion.
As mentioned previously, we focus on the Ethics of AI rather than Machine Ethics. However, further to this, we use the term ethics based on guidelines within applied ethics, particularly in relation to machine understanding. In Döring et al. (2011), the principles of beneficence, non-maleficence, autonomy, and justice are set out as being fundamental considerations for those working in AI. Although this is particular to emotionally aware systems, we consider that such principles are relevant across AI research. Of particular relevance to this contribution, is autonomy, i.e., a duty for systems to avoid interference, and respect an individual's capacity for decision-making. This principle impacts upon both data representation and infrastructure choices (e.g., centralized or decentralized).
We consistently refer to the term bias throughout our contribution. First introduced to machine learning by Mitchell (1980), we typically discuss statistical biases, unless otherwise stated, which may include absolute or relative biases. To be more specific, we focus closely on data in this contribution, and therefore dominantly refer to selection-bias. Selectionbias stems in part from prejudice-based biases (Stark, 2015). However, selection-bias falls within statistical biases as it is a consequence of conscious (hence prejudice) or unconscious data FIGURE 1 | A simplified overview of a typical centralized (left) and decentralized (right) network infrastructure. In the right figure individuals choose the modality to share (as indicated by circle, square, and triangle icons), and users in the network have agency in how their data is used. In the left figure, the AI is essentially a black-box, and users make all modalities of data available to all components of the AI infrastructure.
selection. Selection-bias is particularly relevant to AI given that real randomization (or diverse representation) of data is not always possible.
As a critical aspect of our contribution, relating to the mitigation of bias, through a more ethical approach to data infrustructure, we consistently refer to decentralized AI. A broad definition of decentralization is the distribution of power moving away from central authorities. In the context of AI, when discussing decentralization, we refer to decentralized architectures which allow for this type of distribution, in regards to data sourcing, management and analysis. We do touch on literature relating to blockchain, which is a well-known decentralized approach. However, the term is utilized here more generally and is not exclusive to the blockchain.

BACKGROUND: BIAS AND DECENTRALIZATION IN AI
Funding and global research efforts in the field of AI have increased in the last decade, particularly in the areas of health, transportation, and communication (Mou, 2019). Along with this increase has come a rise in ethical demands related to Big Data (Herschel and Miori, 2017). Although true Big Data is said to need Veracity, the reality of this is sometimes different, with large-scale data often showing particular biases toward clustered demographics (Price and Ball, 2014). As a result, terms, such as Machine Learning Fairness-promoted initially by Google Inc. 2 -is now regularly referred to in an endeavor to build trust and show ethical sensitivity (Mehrabi et al., 2019). In this regard, IBM released their AI Explainability 360 Toolkit 3 in which the overarching goal appears to be improving trust in AI, through more deeply researching machine learning biases, as it pertains to the research areas of fairness, robustness and explainability.
Three common forms of bias are discussed concerning AI, i.e., interaction-bias, latent-bias, and selection-bias. Selection-bias occurs when the data used within a paradigm is selected with bias, leading to misrepresentation rather than generalization. In particular, researchers are repeatedly finding bias in regards to gender (Gao and Ai, 2009). Wang et al. (2019a) found for example that models tend to have a bias toward a particular gender even when a dataset is balanced-which could point to lower level architecture-based biases (Koene, 2017). Selection-bias is essential to combat when referring to models developed for human interaction. Based on data decision making, a bias can propagate through system architectures, leading to lower accuracy on a generalized population. Lack of generalization is particularly problematic for domains, such as health, where this may result in a breach of patient safety (Challen et al., 2019).
Furthermore, the evaluation of fairness in machine learning is another prominent topic, highlighted as a machine learning consideration in Hutchinson and Mitchell (2019). Additionally, researchers propose fairness metrics for evaluating the bias which is inherent to a model (Friedler et al., 2019), including the Disparate Impact or Demographic Parity Constraint (DPC). DPC groups underprivileged classes and compares them to privileged classes as a single group. Similarly, there are novel architectures which mitigate bias through prioritization of minority samples, and the authors of this approach suggest that there is an improvement in generalized fairness (Lohia et al., 2019).
A core contributing factor to bias in AI is the management of data. Current AI networking is based on centralized infrastructure (cf. Figure 1), where individuals present a unified data source to a central server. This centralization approach not only limits privacy but also creates a homogeneous representation, which is less characteristic of the individual interacting (Sueur et al., 2012).
Decentralization in AI was initially coined as a term to describe "autonomous agents in a multi-agents world" (Miiller, 1990), and researchers have proposed decentralization for large AI architectures e.g., integrating machine learning with a Peer-to-peer style blockchain approach Zheng et al., 2018] to improve fairness and bias (Barclay et al., 2018). In this architecture, collaborative incentives are offered to the network users and approaches allow for improved identityrepresentation, as well as more control in regards to data-usage, resulting in more freedom and higher privacy. Furthermore, a decentralized network may inherently be more ethical as more individuals are interacting with and refining the network with agency (Montes and Goertzel, 2019).
For individuals interfacing with AI, privacy is a concern (Montes and Goertzel, 2019). Improving privacy is a core advantage of decentralized data approaches (Daneshgar et al., 2019). In a centralized approach, anonymization processes exist (e.g., that which are enforced by GDPR), although it is unclear how this is consistently applied. To this end, identification of a participant in the data source may not be needed, yet, unique aspects of their character (e.g., how they pronounce a particular word), are still easily identified (Regan and Jesse, 2019).
There are multiple organizations and corporations which focus on the benefits of decentralization, including Effect.AI and SingularityNET 4 Such organizations promote benefits including "diverse ecosystems" and "knowledge sharing." The Decentralized AI Alliance 5 is another organization which integrates AI and blockchain, promoting collaborative problemsolving. In general, the term decentralization comes not only from technical network logistic but from philosophical "transhuman" ideologies (Smith, 2019). In regards to the latter, decentralization promotes the improvement of human-wellbeing through democratical interfacing with technology Goertzel (2007). This democratic view is one aspect of decentralization that aids in the reduction of AI bias (Singh, 2018).
Similarly, there are organizations which focus primarily on the challenge of bias in AI, from many viewpoints including race, gender, age, and disability 6 , most of which implement responsible research and innovation (RRI). When applying RRI to the AI community, the aim is to encourage researchers to anticipate and analyse potential risks of their network, and ensure that the development of AI is socially acceptable, needed, and sustainable (Stahl and Wright, 2018). Biases are an essential aspect of AI RRI (Fussel, 2017), as poor identity-representation has dire consequences for real-world models (Zliobaite, 2015).

METHODOLOGY: ETHICAL DATA CONSIDERATIONS
There are an array of concerns relating to the ethics of AI, including, joblessness, inequality, security, and prejudices (Hagendorff, 2019). With this in mind, academic and industry-based research groups are providing tools to tackle these ethical concerns (cf . Table 1), mainly based on four key areas. In this section, we introduce and conceptually discuss these four ethical considerations-auditing, benchmarking, confidence and trust and explainability and interpretabilitychosen, due to their prominence within the AI community. As well as this, these four aspects, each have a pivotal impact on data representation, and an inherent relation to data infrastructures. An overview of a typical machine learning workflow with these four considerations highlighted based on their position in time is given in Figure 2. To this end, herein, we first define our four considerations more concretely, followed by a description of specific ethical concerns ([±]) which relate to them.

Auditing
In the context of AI data, auditing is not dissimilar to research domains, such as economics. An auditor regularly checks aspects of the system, including the data validity itself. For example Fernández and Fernández (2019) propose an AI-based recruiting systems-in which the candidate's data is validated by a manual (i.e., human) auditor. In Figure 2 we have assigned auditing to every aspect of the AI workflow, although it is commonly only integrated during earlier development stages.
[±] Auditing is integral as acquisition scales up to Big Data. The process of managing what Schembera and Durán (2020), describes as "tangible data" can be extremely time-consuming and costly for those involved and human or machine error can propagate, resulting in biases or leading to mostly unusable data (L'heureux et al., 2017). On the other side, is the auditing of "dark data." This data type is estimated to be 90% (Johnson, 2015) of all stored data, and is largely unknown to the user. The literature currently focuses on auditing tangible data, as yet there is less attention for dark data (Trajanov et al., 2018).

Benchmarking
In machine learning, benchmarking is the process of evaluating novel approaches against well-establish approaches or databases of the same task. To this end, it often comes at a later stage during the AI workflow (cf. Figure 2). In the computer vision domain, this has been particularly successful in pushing forward developments (Westphal et al., 2019), with data sets, such as MNIST (LeCun and Cortes, 2010) or CIFAR-10 (Krizhevsky et al., 2009), continuously benchmarked against in both an  academic and industry setting. Pre-trained networks are another benchmarking tool. Networks, such as imageNet (Simon et al., 2016) are well-known and consistently applied, given the quantity of data and promising results (Wang et al., 2019d).
[±] Multimodal analysis is becoming more ubiquitous in machine learning (Stappen et al., 2020), due to well-known and longstanding advantages (Johnston et al., 1997). When datasets are multimodal benchmarking improvements accurately becomes complex (Liu et al., 2017), and aspects, such as modality missmatches are common (Zhang and Hua, 2015). Additionally, given the rapid developments in machine learning approaches, outdated methods may be held as benchmarks for longer than is scientifically meaningful.

Confidence and Trust
In AI data, the terms confidence and trust are applied to ensure reliability, i.e., having confidence in the data results in deeper trust (Arnold et al., 2019). In this context, trust is a qualitative term, and although confidence can fall into these interpretations relating to enhanced moral understanding (Blass, 2018), the term confidence typically refers to a quantifiable measure to base trust on (Zhang et al., 2001;Keren et al., 2018).
[±] Not providing an overall confidence for resulting predictions, can result in a substantial risk to the user (Ikuta et al., 2003), i.e., if a trained network has an inherent bias, a confidence measure improve the transparency of this. Furthermore, to increase trust in AI, developers are attempting to replicate human-like characteristics, e.g., how robots walk (Nikolova et al., 2018). Adequately reproducing such characteristics, requires substantial data sources from refined demographics. This concern falls primarily into Machine Ethics, with the need for binary gender identifications , and the societal effect of doing so challenged (Jørgensen et al., 2018).

Explainability and Interpretability
Often referred to as XAI (eXplainable AI) and arguably at the core of the ethical debate in the field of AI is explanabilty and interpretability. These terms are synonymous for the need to understand algorithms' decision making (Molnar, 2019;Tjoa and Guan, 2019). However, a distinction can be made, interpretability being methods for better understanding a machine learning architecture or data source (i.e., the how), and explainability being methods for understanding why particular decision were made.
[±] A surge in machine learning research, has come from international challenges (Schuller et al., 2013;Ringeval et al., 2019)-driving improvements in accuracy across multiple machine learning domains (Meer et al., 2000). However, this fast-paced environment often leaves less time for interpreting how particular features may have explicitly impacted a result, or for an explanation of a models decision-making process. Without this, the meaning of any result is less easy to substantiate (Vellido et al., 2012).

DISCUSSION: REPRESENTATION AND INFRASTRUCTURE
Having defined our four key consideration more concretely, we now discuss them more closely with representation (w.r.t., bias) and infrastructure of AI data in mind. Where meaningful, we highlight technical approaches which are implemented to reduce the aforementioned ethical concerns.

Auditing
There are many methods being developed to make collecting and annotating data in an automatic way possible, including data mining of web-based images (Zafar et al., 2019), and active learning (AL) for semi-automatic labeling (Wang et al., 2019c). For data tagging by autonomous agents, some have shown concerns that making agents responsible for this, may lead to incorrect tagging caused by an initial human error. A concern which becomes more problematic given the now large quantities of child viewers, who may be suggested inappropriate content (Papadamou et al., 2019). Further to this when annotating data, one ethical issue which can propagate selectionbias is poorly balanced manual vs. automatic annotations. In other words, if automatic annotation procedures learn false aspects early on, these may then be replicated (Rothwell et al., 2015). In an AL paradigm (Ayache and Quénot, 2008), an oracle (i.e., expert auditor) is kept in the loop, and where the AL model is uncertain at a particular level of confidence, the oracle must provide the label (Settles et al., 2008). In the case of specialist domains, such as bird sound classification, having such an expert is crucial, as variances in the audio signal can be quite slight (Qian et al., 2017).
Within a larger decentralized network, utilizing auditors allows for a democratic style of data management. Blockchain AI networks, for example, run in a peer-to-peer (P2P) fashion, meaning that no changes can be made to the system without the agreement of all others in the network. In a P2P network, there is an incentive for individual participation in the auditing process (e.g., an improved overall experience) (Dinh and Thai, 2018). However, the realization of auditing in AI does lead to some technical challenges in regards to public verification of sensitive data (Diakopoulos and Friedler, 2017), as well as making the AI only a partial reduction of human time-cost. Nevertheless, the need for auditing in AI has been highlighted consistently in the literature as a bias mitigating approach (Saleiro et al., 2018)

Benchmarking
It has been noted in many domains of research that benchmarking and therefore generalizing against a well-established organization, may result in the continued propagation of poor standards concerning historical biases (Denrell, 2005). Survey-based evaluations of the state-of-the-art modalities and baselines results are one resource to help mitigate this issue (Liu et al., 2011;Cummins et al., 2018). However, constant updates to benchmarks should be made, updating both techniques for acquisition and methods for setting baselines. Although there is no rule of thumb in this case, it is generally accepted in machine learning that benchmarking against resources that are no longer considered to be state-of-the-art will not bring valid results. Furthermore, in the realm of human-data, and specifically within the European Union, there is often a limited time that data can be stored (The-European-Commission, 2019). In this way, not only will benchmarked data sets become outdated in terms of techniques, but it is unethical to utilize such data, as reproducibility may not be possible.
Of note, a considerable contribution for ethics-based benchmarking is the aforementioned open-source IBM AI Explainability 360 Toolkit, in which one aspect is the Adversarial Robustness 360 Toolbox. This toolbox provides state-of-the-art paradigms for adversarial attacks (i.e., subtle alterations to data), and allows researchers to benchmark their approaches in a controlled environment to allow for more easy interpretation of possible network issues.

Confidence and Trust
Given the general fear that members of the public have for AI-mostly attributed to false depictions in movies and literature -improving confidence and trust in AI is now at the forefront for many corporations. To this end, researchers and corporations continually introduce state-of-the-art aids for tackling famous AI problems, such as the IBM AI Fairness 360 Toolkit. As well as this, to improve trust groups, such as "IBM Building Trust in AI" 7 , make this their specific focus. In this particular group, developing human-like aspects is given a priority, as research has shown that humans trust the general capability of more human-like representations over purely mechanical ones (Charalambous et al., 2016). However, the well-known uncanny valley (which refers to familiarity and likeability, concerning human-likeness) suggests that data-driven representations requiring trust should be very-near humanlike (Mori et al., 2012), and action may result in biased binary representations, which may be problematic in terms of identity politics (Jørgensen et al., 2018).
Another effort in improving trust comes from blockchain. Blockchain is a specific decentralized approach known as a distributed digital ledger, in which transactions can only be altered with the specific agreement of subsequent (connected) blocks (Zheng et al., 2018). Blockchain is said to offer deeper trust for a user within a network, due to the specific need for collaboration (Mathews et al., 2017). This approach offers further accountability, as decisions, or alterations are agreed upon by those within the network. More specifically, trust is established through algorithms known as consensus algorithms (Lee, 2002).
As mentioned, one quantifiable measure to build on trust are confidence measures, sometimes referred to as uncertainty measures i.e., those applied in a semi-automated labeling paradigm. A confidence measure evaluates the accuracy of a model's predictions against a ground truth or set of weights and provides a metric of confidence in the resulting prediction (Jha et al., 2019). Herein, we follow this definition for confidence as a measure, i.e., how accurate is the current system prediction, as a means of understanding any risk (Duncan, 2015). This definition allows researchers to have a margin of error and can be a crucial aspect of the health domain to avoid false-positives (Bechar et al., 2017).
Given the "black-box" nature of deep learning, there have been numerous approaches to quantifying confidence (Kendall and Cipolla, 2016;Keren et al., 2018). One popular procedure for measuring confidence is the Monte Carlo dropout. In this approach, several iterations are made, each time "dropping" a portion of the network, and calculating confidence or uncertainty based on the variance of each prediction (Gal and Ghahramani, 2016).
As an additional note, data-reliability is a term often referred to in regards to both confidence and trust. Typically this is the process of statistically representing the significance of any findings from the database in a well-established scientific fashion, particularly considering the context of the domain it is targeted toward (Morgan and Waring, 2004). Statistical tests, such as the p-value, which is used across research domains, including machine learning, remains controversial. A p-value, states the strength (significance) of evidence provided and suffers from the "dancing p-value phenomena" Cumming (2013). This phenomenon essentially shows that in a more real-world setting the p-value can range (within the same experimental settings) from <0.001 to 0.5, i.e., from very significant to not significant all. Given this limitation, the researcher may present a biased experiment, in an endeavor to report a significant result. This limitation of the p-value, amongst other statistical tests, has gained criticism in recent years, due to their extensive misuse by the machine learning community (Vidgen and Yasseri, 2016).

Explainability and Interpretability
Researchers continue to work towards more accurately understanding the decisions made by deep networks (Huszár, 2015;Rai, 2020). Machine learning models must be interpretable and offer a clear use-case. At the core of this, data itself in such systems should also be explainable i.e., designed data acquisition, with plausible goals. Machine learning is a pattern recognition task, and due to this visualization of data is one way to help with detailing both interpretability and explainability of a system by (1) better understanding the feature space, and (2) better understanding possible choices. In regards to the bias in AI, visualization of data-points allows for a more easily determined observation of any class dominance. Clustering is a particular pre-processing step applied in Big Data-based deep learning (Samek et al., 2017). Popular algorithms which apply this type of visualization include t-distributed stochastic neighbor embedding (t-SNE) (Zeiler and Fergus, 2014) and Laplacian Eigenmaps (Schütt et al., 2019). More recently, there has been a surge in approaches for visualizing attention over data points (Guo et al., 2019). These approaches are particularly promising as they show visually the areas of activation which are learnt most consistently for each class by a network (Wang et al., 2019b), therefore highlighting areas of bias more easily, and improving communication methods to those outside the field.
To this end, decentralization with integrated blockchain is one approach which has been noted as improving interpretability, mainly as data is often-publicly accessible (Dinh and Thai, 2018). For example, where bias begins to form, the diversity of modalities and ease in identification means that individual blocks can be excluded entirely from a network to meet a more accurate representation (Dai et al., 2019).

FUTURE DIRECTIONS
Due in part to the ethics-based commitments by some of the larger AI companies, we see from this review that, there is momentum toward a more ethical AI future. However, interdisciplinarity in AI research is one aspect which requires more attention. To the best of the authors' knowledge, most public forums (particularly those based on a centralized infrastructure) come from a mono-domain viewpoint (e.g., engineering). Incorporating multiple disciplines in the discussion appears to be more prominent with those promoting decentralized AI.
Interdisciplinary will not only improve implementation of the four ethical consideration described herein, but has been shown to be a necessary step forward for the next AI phase of Artificial General Intelligence (AGI), proposed by the decentralized community (Goertzel and Pennachin, 2007). Interdisiplinarity is particularly of value as infrastructures developed in this way more easily tackle ethical concerns relating to; (i) integration, (ii) selection-bias, and (iii) trust.
Seamless integration of AI is necessary for its success and adoption by the general public. Aspects including cultural and environmental impact need to be considered, and various experts should provide knowledge on the target area. For example, the synthesized voice of bus announcements not representing the community to which it speaks may have a negative impact on those communities, and a closer analysis of the voice that best represents that community would be more ethically considerate. In this way, working alongside linguists and sociologists may aid development.
Similarly, from our literature overview, we observe that knowledge of selection-bias often requires contributions from experts with non-technical backgrounds, and an approach for facilitating discussion between fields of research would be a valuable next step. For example, within the machine learning community, techniques, such as few-shot learning are receiving more attention in recent years (Wang and Yao, 2019), however, perceptual-based biases pose difficulties for such approaches (Azad et al., 2020), and discussion from experts of the targeted domains may help understand the bias at an earlier stage. Despite this, communication between fields speaking different "languages" (i.e., anthropology and engineering), is a challenge in itself, which should be addressed by the community. Furthermore, due to historical stereotypes, AI continues to lack in trust by the general user. Users who without an understanding of the vocabulary of the field, may not be able to grasp the concept of such networks. Through a better collaboration with various academic researchers, communicating AI to the general public may also see an improvement, which in turn will help to build trust and improve wellbeing of the user during AI interaction.

CONCLUSION
The themes of data representation and infrastructure as they pertain to selection-bias and decentralization in AI algorithms have been discussed throughout this contribution. Within these discussion points, we have highlighted four key consideration; auditing, benchmarking, confidence and trust, and explainability and interpretability to be taken into account when handling AI data more ethically.
From our observation, we conclude that for all of the four considerations, issues which may stem from multimodal approaches should be treated cautiously. In other words, relating to auditing, there should be standards for each modality monitored, as this follows through into the ability for accurate benchmarking. In this same way, although the literature may argue this, confidence and trust come from diverse representations of human data, which in turn are more explainable to the general public due to its inherent human-like attributes.
With this in mind, we see that efforts are being made, for fully audited, benchmarkable, confident, trustworthy, explainable and interpretable machine learning approaches. However, standardization for the inclusion of all of these aspects is still needed. Furthermore, with the inclusion of multiple members who take equal responsibility, decentralization may enable the ethical aspects highlighted herein. We see that through socialmedia (which is in some sense a decentralized network for communication) group morality is developed. Opinions of a political nature, for example, are highlighted, and any prejudices or general wrongdoing is often shunned and which can have enormous impact on business (Radzik et al., 2020). In this way, a more transparent and open platform makes masking potential network biases a challenge.

AUTHOR CONTRIBUTIONS
AB: literature analysis, manuscript preparation, editing, and drafting manuscript. BS: drafting manuscript and manuscript editing. All authors revised, developed, read, and approved the final manuscript.